Reversal of the Concreteness Effect for Verbs in Patients With Semantic Dementia

Transcription

1 Neuropsychology Copyright 2007 by the American Psychological Association 2007, Vol. 21, No. 1, /07/$12.00 DOI: / Reversal of the Concreteness Effect for Verbs in Patients With Semantic Dementia Hyon-Ah Yi University of Pennsylvania School of Medicine and Keimyung University School of Medicine Peachie Moore and Murray Grossman University of Pennsylvania School of Medicine The authors assessed comprehension of carefully matched classes of words, manipulating grammatical subcategory (noun and verb) and semantic (concrete and abstract) characteristics for participants with semantic dementia (SD) or probable Alzheimer s disease (AD). Participants selected the best of four words that matched a verbal description. Participants with AD or SD were significantly impaired with verbs compared with nouns. Moreover, participants with SD showed significantly greater difficulty with motion verbs compared to cognition verbs. The authors argue that two factors contribute to the difficulty with motion verbs for patients with SD. First, the verb semantic network is very poorly organized relative to the noun semantic network, leaving verbs more vulnerable to a progressive neurodegenerative disease. Second, visual feature knowledge is degraded in patients with SD because of the anatomic distribution of the disease in visual association cortex, causing relatively greater difficulty for concrete verbs compared to abstract verbs. Keywords: semantic dementia, frontotemporal dementia, Alzheimer s disease, semantic, verb Semantic memory refers to memory for a shared pool of knowledge about the world that, unlike episodic memory, is common to many individuals. It includes the long-term representations of objects, actions, ideas, and thought processes that allow people to communicate meaningfully with others. Most categories of knowledge investigated in previous research have been confined to concrete nouns. Far fewer studies have assessed differences between nouns and verbs, or between concrete and abstract concepts. In the present study, we investigated category-specific effects for grammatical (noun vs. verb) and semantic (concrete vs. abstract) characteristics of words in neurodegenerative diseases associated with semantic difficulty. Crucial insights into the nature of semantic memory have come from seminal observations of category-specific difficulty. Perhaps the most common semantic memory deficit addresses the comparison of natural kinds and manufactured artifacts (Lu et al., 2002; Silveri & Gainotti, 1988; Warrington, 1975; Warrington & Shallice, 1984). Cross-sectional and longitudinal studies of categoryspecific semantic memory in patients with probable Alzheimer s disease (AD) have shown a deficit for natural kinds (Garrard, Patterson, Watson, & Hodges, 1998; Garrard et al., 2001; Silveri, Daniele, Giustolisi, & Gainotti, 1991). Studies of patients with Hyon-Ah Yi, Department of Neurology, University of Pennsylvania School of Medicine, and Department of Neurology, Keimyung University School of Medicine, Daegu, South Korea; Peachie Moore and Murray Grossman, Department of Neurology, University of Pennsylvania School of Medicine. Portions of this work were presented at the annual meeting of the American Academy of Neurology, San Diego, California, April This work was supported in part by National Institutes of Health Grants AG15116, AG17586, and NS Correspondence concerning this article should be addressed to Murray Grossman, Department of Neurology, University of Pennsylvania School of Medicine, 3 W Gates Building, 3400 Spruce Street, Philadelphia, PA [email protected] semantic dementia (SD), a neurodegenerative disease associated with significant semantic difficulty, also have shown an impairment of natural kinds (Lambon-Ralph, Patterson, Garrard, & Hodges, 2003; Warrington, 1975). The most prominent hypothesis forwarded to explain this category-specific effect is based on the notion that conceptual representations of living things and manufactured artifacts depend on visual perceptual and functional information, respectively (Farah & McClelland, 1991; Warrington & Shallice, 1984). Although there is much evidence consistent with this account of category-specific semantic memory, many problems also have been noted. This includes both the role that visual perceptual feature knowledge plays in a deficit for a specific semantic category (Caramazza, 2000; Caramazza & Shelton, 1998) and the inconsistent association of category-specific knowledge with functional magnetic resonance imaging (fmri) studies of visual association cortex and semantic memory (Joseph, 2001). An alternate approach to category-specific semantic memory difficulty invokes a distributed network model (Gonnerman, Andersen, Devlin, Kempler, & Seidenberg, 1997; Moss & Tyler, 2000). According to this view, microscopic damage selectively degrades portions of the network storing semantic knowledge while the remaining intact cortex compensates for degraded knowledge. If a feature of poodle is degraded, for example, this knowledge can be borrowed from the set of neighboring exemplars contributing to the superordinate concept dog (Gonnerman et al., 1997; Rogers et al., 2004). Although elegant computational models provide an existence proof consistent with these distributed accounts, longitudinal studies of AD and SD do not provide strong support for this approach (Garrard et al., 2001; Grossman et al., 2006). The present study takes advantage of a more reliable aspect of semantic memory in AD and SD the concreteness effect. The concreteness effect refers to the common observation of superior performance for concrete words compared with abstract ones. The dual-coding hypothesis (Kounios & Holcomb, 1994; Paivio, 1991) attributes the advantage for concrete words to encoding by both a 9

2 10 YI, MOORE, AND GROSSMAN verbal propositional system and a visual perceptual system, whereas abstract words are encoded only by a verbal propositional system. Studies of patients with AD generally have shown that patients have more difficulty understanding abstract words than concrete words, presumably because of the less redundant and more fragile representation of abstract concepts relative to concrete concepts in the face of a neurodegenerative disease (Kroll & Merves, 1986; Rissenberg & Glanzer, 1987). However, occasional case reports of SD have described a reversal of the concreteness effect (Breedin, Saffran, & Coslett, 1994; Cipolotti & Warrington, 1995; Sirigu, Duhamel, & Poncet, 1991; Warrington, 1975, 1981; Warrington & Shallice, 1984). In these rare descriptions, patients with SD show significantly better performance at identifying and reading abstract words than concrete words. The dual-coding hypothesis has difficulty explaining this phenomenon. A neuroanatomic account suggests instead that knowledge of concrete words may depend in part on visual association cortex in the inferior temporal lobe, whereas abstract words may depend on some other neuroanatomic representation concerned with complex relationships in neural networks, such as prefrontal cortex. Because patients with SD have inferior temporal lobe disease, which is important for visual association cortex, they may be more susceptible to difficulty with concrete concepts and thus more likely to show reversal of the concreteness effect. Semantic memory difficulty in patients with AD, by comparison, tends to be associated with modality neutral association cortex in a posterolateral temporal distribution (Desgranges et al., 1998; Grossman, Koenig, Devita, et al., 2003; Grossman et al., 1997; Harasty, Halliday, Kril, & Code, 1999; Penniello et al., 1995; Prvulovic et al., 2002). In the work described above, researchers have investigated hypotheses concerned with semantic memory using nouns. A handful of studies have shown that verbs are more difficult than nouns for patients with AD or frontotemporal dementia (FTD; Bak, O Donovan, Xuereb, Boniface, & Hodges, 2001; Bushell & Martin, 1997; Cappa et al., 1998; Hillis, Sangjin, & Ken, 2004; Hillis, Tuffiash, & Caramazza, 2002; Parris & Weekes, 2001; K. M. Robinson, Grossman, White-Devine, & D Esposito, 1996; White-Devine, Grossman, Robinson, Onishi, & Biassou, 1996). The fact that this is not a universal finding (Fung et al., 2001; G. Robinson, Rossor, & Cipolotti, 1999; Williamson, Adair, Raymer, & Heilman, 1998) emphasizes that verbs are not inherently more difficult than nouns. Indeed, fmri investigations of grammatical category effects in semantic memory point to the distinct neuroanatomic representations for verbs and nouns (Grossman et al., 2002; Kable, Lease-Spellmeyer, & Chatterjee, 2002; Perani et al., 1999; Tyler, Russell, Fadili, & Moss, 2001), although the precise neuroanatomic locus for these word classes appears to be quite variable. Examination of grammatical subcategory nevertheless may be helpful in understanding the basis for semantic memory difficulty and the concreteness effect in patients with AD or SD. Consider in this context the hypotheses forwarded to explain differences between verbs and nouns. One hypothesis focuses on the possibility that extensive resources are needed to process all of the grammatical, thematic, and semantic information represented in verbs relative to nouns, although experimental investigations have not supported this claim (Rhee, Antiquena, & Grossman, 2001; K. M. Robinson et al., 1996; White-Devine et al., 1996). Another possibility emphasizes the importance of lexical frequency in retrieving category-specific semantic knowledge. Bird, Lambon-Ralph, Patterson, and Hodges (2000) examined word frequency in their analysis of word production during descriptions of the Cookie Theft Scene. They suggested that the noun production impairment, seen early in the course of SD, is due to selective difficulty with low frequency nouns. This frequency-sensitive effect was not seen for verbs. A final hypothesis is that differences between verbs and nouns are due in part to the distinct semantic network characteristics of these word classes. Nouns thus have a well-organized, deeply layered hierarchical structure that is supported by a rich, distributed network, as noted above (Gonnerman et al., 1997; Rogers et al., 2004). In contrast, verbs have a poorly organized, relatively shallow hierarchical structure that provides little support for the verb semantic network (Miller & Fellbaum, 1991). Because the semantic network for verbs is much less organized, feature representations and the feature-sharing compensatory mechanism of a semantic network may be much more vulnerable for verbs than nouns in the face of neurodegenerative diseases, such as SD or AD. We tested these semantic (concrete vs. abstract) and grammatical (noun vs. verb) word category effects in the present study. We hypothesized that two interacting components contribute to lexical comprehension difficulty in patients with SD or AD. First, we hypothesized that verb semantic memory would be more vulnerable to the degradation of feature knowledge than noun semantic memory in patients with neurodegenerative diseases because of the poorly structured semantic network for verbs relative to nouns. Although nouns may be able to compensate for degraded knowledge by borrowing features shared with neighboring concepts, this compensatory mechanism may be less successful for verbs because their poorly structured semantic network makes it difficult to identify overlapping concepts that share features. Second, knowledge for concrete verbs was expected to be particularly vulnerable in patients with SD because of the anatomic distribution of their disease in visual association cortex. Despite the compensatory mechanism afforded by the well-structured noun semantic network, degraded visual perceptual feature knowledge may also have some impact on concrete nouns. Participants Method We studied 41 participants with a neurodegenerative disease thought to compromise semantic memory, including 29 participants with probable AD and 12 participants with SD. These participants were compared with 17 healthy older controls. Participant diagnosis was rendered by a neurologist with experience studying neurodegenerative conditions according to published criteria: From the National Institute of Neurological and Communicative Diseases and Stroke Alzheimer s Disease and Related Disorders Association for AD (McKhann et al., 1984), and a modification of criteria described by Neary et al. (1998) for SD (Grossman & Ash, 2004). These diagnoses were confirmed by evaluations of the written clinical record in a consensus conference. All participants were right-handed native speakers of English recruited from the Department of Neurology at the Hospital of the University of Pennsylvania (Philadelphia). At the initial visit, they underwent a detailed clinical interview, a neurological examination, and a complete evaluation of mental status. Laboratory tests and structural brain imaging studies excluded other causes of dementia such as vascular disease, primary psychiatric disorders, and medical illnesses or metabolic conditions that may have caused a cognitive deficit. The demographic

3 REVERSED CONCRETENESS EFFECT FOR VERBS 11 features are summarized in Table 1. The age of participants with SD was significantly younger than the participants with AD, t(39) 3.38, p.05, as we would expect from demographic studies of these patients groups (Neary et al., 1998), and was significantly younger than healthy control participants, t(26) 5.98, p.05. The AD and SD groups were otherwise matched in their dementia severity, according to the Mini-Mental State Examination (Folstein, Folstein, & McHugh, 1975) score, t(39) 0.002, ns, and education, t(39) 0.15, ns. The controls were significantly less impaired on the Mini-Mental State Examination than participants with AD, t(44) 6.72, p.05, and participants with SD, t(27) 6.11, p.05, whereas the controls were matched to both participants with AD, t(43) 2.16, ns, and SD, t(26) 1.68, ns, in their education. To verify participants difficulty, we administered the Boston Naming Test (Goodglass, Kaplan, & Weintraub, 1983) and a category-naming fluency test for animals. The performance of each group is also summarized in Table 1 as z scores relative to the age- and education-matched healthy seniors. Materials Our core measure was a multiple-choice, naming-to-description task. The target words were taken from two different major grammatical categories: nouns and verbs. Forty verbs and forty nouns were used. The descriptions being judged by participants were adopted from Merriam- Webster s International Dictionary (3rd. ed.; 1986) and were modified into colloquial English. The target words are listed in the Appendix. We selected stimuli that were present in this frequency count as only verbs or only nouns, or that were much more frequent ( 7:1 ratio) as one grammatical category relative to the other. A form class-sensitive frequency count (Francis & Kucera, 1982) was used to verify that the targets occurred more frequently in one grammatical form class than in the other. There was no significant difference between the frequency of target verbs and target nouns on the basis of semantic categories: Concrete nouns versus motion verbs, t(38) 2.01, ns, and abstract nouns versus cognition verbs, t(38) 1.41, ns, did not differ. We also measured the familiarity of target nouns and verbs. Ten native speakers of English were asked to complete a questionnaire in which they judged the familiarity of the target words on a scale ranging from 1 (not familiar) to 5(very familiar). The familiarity of the verb targets was matched to that of the noun targets, t(18) 0.33, ns. Each grammatical category had equal numbers of concrete and abstract words, including concrete and abstract nouns, and motion and cognition verbs. There were 20 verbs of motion and 20 verbs of cognition. Verbs of motion refer to words that are related to object manipulation and describe an action. Verbs of cognition are words that are related to psychological or mental states, having little association with motion. The frequency of occurrence of verbs of motion was matched to that of verbs of cognition, t(38) 0.26, ns. We also used 20 concrete nouns and 20 abstract nouns. Concrete nouns are words that can be readily observed with the senses and possess physical attributes. Abstract nouns include concepts, such as legal terms and mental states, that are not readily accessible by the senses. The frequency of occurrence of concrete nouns was matched to that of abstract nouns, t(38) 0.46, ns. We designed the task to use a multiple-choice presentation format so that the role of participants difficulty with naming production or lexical retrieval could be minimized. Each target word thus was presented to the participants together with three foils. These foils are also provided in the Appendix. For the verbs, there were the following three types of foils: (a) reverse-role foil, in which the agent and recipient exchange roles, such as buy for the target word sell ; (b) semantically-related-to-target-word (SRTW) foil, in which a foil shares many semantic attributes but does not exchange thematic roles, such as convey for the target word sell ; and (c) semantically-related-to-reverse-role (SRRR) foil, in which a foil has a weaker semantic relation with the target, such as steal for the target word sell. Similarly, for the nouns, there were three types of foils: (a) oppositemeaning-to-target-word (OMTW) foil, in which a foil expresses the opposite meaning of the target word, such as conflict for the target word agreement ; (b) SRTW foil, in which a foil is semantically related with the target word, such as gratitude for the target word agreement ; and (c) semantically-related-to-opposite-meaning-to-target-word (SROMTW) foil, a choice that has a weaker relation to the target, such as battle for the target agreement. To diminish the possibility that choices could be affected by frequency, familiarity, and associativity between the target and the foils, we measured these factors for foils. We found no statistically significant differences for frequency and familiarity across the foils for noun and verb target categories (all ps.05). We assessed the associativity strength between the target and each of its foils to minimize the likelihood that this factor affected participants choices. A questionnaire assessing the association between word pairs was administered to 8 healthy native speakers of English. The associativity strength was scored on a scale ranging from 1 (not associated) to 5(very highly associated). We did not find a statistically significant difference across categories of target words ( p.05 for all pairs of word categories). Each description was in the following format: (superordinate semantic category), (description of important features). For example, the description for sell was an action involving the exchange of an object for money. We presented these in writing to diminish the possibility that working memory limitations would interfere with the performance of patients with AD. Each description was no longer than one line, with a word range from 2 to 12 words. We analyzed the frequencies of each content word in each description and compared these across the classes of target word Table 1 Demographic Features of Participants Alzheimer s disease Semantic dementia Healthy senior controls Variable M SD M SD M SD n Women 17 7 Men 12 5 Age at examination (years) Education (years) MMSE scores (max 30) Boston naming test a 3.6 b Category fluency test (animal) a 2.6 b Note. MMSE Mini-Mental State Examination. a These results correspond to z score, relative to 25 age- and education-matched controls. b n 26.

4 12 YI, MOORE, AND GROSSMAN categories. We did not find any significant difference between the mean frequencies of description for nouns and verbs, t(78) 1.32, ns. Moreover, we did not find a difference for the number of words within the descriptions. The analyses based on nouns versus verbs, t(78) 0.65, ns, and concrete versus abstract words, t(78) 1.53, ns, were not significant. The descriptions and the associated choices were presented in black Arial narrow font at 30-point size printed on 8.5 in. (21.59 cm) 11 in. (27.94 cm) white paper bound in a booklet. The description was placed at the top of the page, with the four written choices vertically arrayed in a random order underneath the description. We also developed a parallel, multiple-choice recognition naming procedure using color video film for the concrete nouns and motion verbs. The videos illustrated the target actions or objects. This was verified in pilot testing with healthy young adults who named the depicted actions and objects accurately. Each video clip was about 3 s in duration. Procedure The task specifically probed each participant s ability to select the word that was best described by the description. We read the description and the four choices aloud while they were being read silently by the participant to minimize the possibility that impaired reading could reduce performance. This oral presentation was repeated for any items at the participant s request. The need for written presentation to patients with AD and oral presentation to patients with SD limited our ability to collect reaction time data. Verbs and nouns were presented in a blocked manner, and the order of administration of these blocks was random. Within each block, concrete and abstract stimuli were ordered randomly. Before the experiment, we presented four practice items to each participant. This illustrated the nature of the materials and familiarized participants with stimulus presentation and modality of response. Participants were encouraged to ask questions about the procedure, and the experimenter gave feedback during practice so that the participants understood the nature of a correct response. The videos were presented with the same choices as the description, arrayed vertically in a random order on a page. The video illustrating each target was presented twice before the participant selected the word that went best with the video. Repetition of the video was offered at the request of the participant, and the choices were read aloud to the participant while the participant read them silently. The videos were presented in a different session than the verbal descriptions, separated by about 1 month. The verbal and video sessions were randomly ordered. Because of intercurrent illnesses, scheduling conflicts, and technical difficulties, data for videos were available for 23 participants with AD and 10 participants with SD. A 3 (group: control, AD, SD) 2 (grammatical category: verb, noun) 2 (semantic category: concrete, abstract) repeated measures ANOVA was performed to assess differences in group response patterns to the grammatical and semantic categories. We found a significant main effect for group, F(1, 54) 18.42, p.001. Post hoc analyses with Tukey s honestly significant difference revealed significantly poorer performance in participants with AD and in participants with SD relative to controls ( p.05). There was also a significant main effect of grammatical category, F(1, 54) 83.11, p.001, but there was not a significant main effect of semantic category, F(1, 54) 0.00, ns. We also found a significant two-way interaction for Group Grammatical Category, F(2, 54) 8.56, p.05, and a significant three-way interaction for Group Grammatical Category Semantic Category, F(2, 54) 4.75, p.05. Mean performance for each set of stimuli in each group is summarized in Figure 1. We describe below additional analyses to understand performance within each group. Participants With AD The pattern of judgments in AD showed more difficulty with verbs than nouns, and more difficulty with abstract nouns than concrete nouns. Thus, participants with AD had worse performance judging verbs than nouns, t(27) 7.89, p.001. We also found that participants with AD had greater difficulty judging abstract nouns than concrete nouns, t(27) 2.70, p.05. However, there was no significant difference between cognition verbs and motion verbs, t(28) 0.08, ns. Analyses of individual patient performance profiles also showed more difficulty with verbs than nouns, although the concreteness effect was less evident. Inspection of z scores showed that 23 of 29 participants with AD were significantly impaired on judging verbs, and an analysis of raw scores showed the effect of worse performance with verbs than nouns in 27 of 28 participants with AD. Greater difficulty with abstract nouns than concrete nouns was present in 13 of 28 participants with AD. Of the 29 participants with AD, 17 revealed more difficulty with cognition verbs than motion verbs. Participants With SD Participants with SD also showed greater difficulty with verbs relative to nouns, but they demonstrated reversal of the concrete- Statistical Analyses Parametric statistical analyses, such as analysis of variance (ANOVA) and t tests, were performed to contrast group performance and to compare conditions within each group. We also analyzed the individual performance profiles of participants with AD and participants with SD using z-score analyses based on control participants performance. We set our threshold for statistical significance at z 2.32, equivalent to p.01 (two-tailed). When control participants performance did not allow us to perform a z-score analysis because of limited or skewed variability, we examined the raw score of individual participants. Results Figure 1. Mean accuracy of judging verbs and nouns that were concrete and abstract. Error bars represent standard deviations.

5 REVERSED CONCRETENESS EFFECT FOR VERBS 13 ness effect with more difficulty judging motion verbs than cognition verbs. Specifically, participants with SD had poorer performance with verbs than nouns, t(11) 4.77, p.001. Moreover, performance with verbs of motion was significantly worse than with verbs of cognition, t(11) 2.36, p.05. A comparison between AD and SD showed that verb judgments were not worse overall in participants with SD than in participants with AD, t(39) 1.11, ns. However, participants with SD performed significantly worse than participants with AD for verbs of motion, t(39) 2.03, p.05, although participants with AD and participants with SD did not differ for verbs of cognition, t(39) 0.00, ns. Noun judgments in participants with SD did not differ statistically from participants with AD. Analyses of individual patient performance profiles confirmed greater difficulty with verbs than nouns in participants with SD and reversal of the concreteness effect for verbs in a majority of these patients. Of the 12 participants with SD, 9 (75%) were significantly impaired in their judgments of verbs, according to z-score analyses, and worse performance with verbs than nouns was seen in 11 of 12 participants with SD. Worse performance with motion verbs than cognition verbs was seen in 9 (75%) of the 12 participants with SD. A significantly larger number of participants with SD than participants with AD showed reversal of the concreteness effect for verbs, 2 (1, N 41) 3.84, p.05. Only 3 (25%) of 12 participants with SD showed reversal of the concreteness effect for nouns, and the frequency of the concreteness effect for nouns did not differ between participants with AD and participants with SD. Reversal of the concreteness effect occurred more commonly for verbs than nouns in participants with SD, 2 (1, N 12) 4.20, p.01. Error Analysis The overall patterns of errors were similar for participants with AD and participants with SD, showing reverse-role errors for verbs and SRTW errors for nouns. As summarized in Table 2, the verb judgment errors of participants with AD included significantly more reverse-role errors than SRTW errors, t(28) 3.62, p.05, and SRRR errors, t(28) 4.53, p.001. For the nouns, the SRTW errors were significantly more frequent than OMTW errors, t(27) 2.16, p.05, and SROMTW errors, t(27) 3.84, p.001. Within the SD group, verb judgment errors included significantly more reverse-role errors than SRTW errors, t(11) 4.72, p.001, and SRRR errors, t(11) 4.13, p.001. For the nouns, the SRTW errors were more frequent than other types of errors: SROMTW errors, t(11) 3.02, p.001; OMTW errors, t(11) 1.41, ns. Frequency Analysis We did not find any effects for the frequency of the target word. Using a median split of the target words based on frequency, we performed a 2 (group: AD, SD) 2 (frequency: upper half, lower half) 2 (grammatical category: verb, noun) 2 (semantic category: concrete, abstract) repeated measures ANOVA to determine whether the frequencies of the target words had an impact on performance of participants with SD and participants with AD. We found a significant main effect of grammatical category, F(1, 38) 78.65, p.001, and a significant two-way interaction for Grammatical Category Semantic Category, F(1, 38) 17.71, p.001, but there was no effect for frequency. We also performed the Frequency Semantic Category Grammatical Category analysis within participants with SD to assess whether frequency had an effect specifically on the presence of the reversal of the concreteness effect. We found that frequency did not affect the performance of participants with SD. Analysis of Video Stimuli We evaluated video stimuli to determine whether verbal descriptions disadvantaged participants performance, but both participants with AD and participants with SD continued to demonstrate greater difficulty judging motion verbs than concrete nouns. Videos were judged accurately by 14 controls (motion: ; concrete: ), t(13) 1.00, ns. This did not differ from accuracy judging descriptions of these verbs and nouns. In the participants with AD, we found that judging videos of motion verbs ( ) was more impaired than judging videos of concrete nouns ( ), t(22) 3.66, p.05. This did not differ from performance with verbal descriptions, according to t tests ( p.10 for both comparisons). In participants with SD, accuracy for concrete nouns was , and accuracy for verbs of motion was Similar to the assessment of written descriptions, judging videos of verbs was significantly less accurate than judging videos of nouns, t(9) 2.94, p.05. A comparison of written-description performance with video stimuli did not reveal Table 2 Frequency of Selecting Foils During Judgment Errors Alzheimer s disease Semantic dementia Type of error M SD M SD Verb Errors related to reverse role Errors that are semantically related to target Errors related to reverse role Noun Errors related to opposite meaning Errors that are semantically related to target Errors that are semantically related to opposite foil

6 14 YI, MOORE, AND GROSSMAN any significant differences in participants with SD, according to t tests ( p.10 for both comparisons). Discussion We examined impaired semantic knowledge in AD and SD two neurodegenerative diseases associated with semantic memory difficulty. We found that both participants with AD and participants with SD were more impaired at recognizing the meanings of verbs than the meanings of nouns on a multiple-choice, namingto-description task. Verb difficulty was confirmed in a multiplechoice assessment of videos. Moreover, we found qualitatively distinct impairment patterns in participants with AD and participants with SD. Participants with SD showed a more severe impairment on motion verbs than cognition verbs, although this was not seen for nouns; participants with AD showed equal difficulty with motion verbs and cognition verbs, and a trend toward greater difficulty with abstract nouns than concrete nouns. We hypothesized that reversal of the concreteness effect for verbs in participants with SD was related in part to two factors: an impoverished hierarchical structure for the verb semantic network that caused greater difficulty for this class of words than nouns, combined with degraded visual perceptual feature knowledge that resulted in greater difficulty with motion verbs than cognition verbs. Most previous studies investigating semantic memory have focused on the representation of nouns, especially concrete nouns. We found that recognition naming is significantly worse for verbs than nouns in a large group of participants with AD. This confirms previous observations of verb comprehension and naming difficulty in participants with AD (Bushell & Martin, 1997; Parris & Weekes, 2001; K. M. Robinson et al., 1996; White-Devine et al., 1996). The present study is the first description of a relative deficit with verbs in participants with SD. Cappa et al. (1998) reported greater difficulty naming with verbs than naming with nouns in patients with FTD. Nonfluent primary progressive aphasics have greater difficulty with verbs than nouns as well (Hillis et al., 2002, 2004). The observation of greater difficulty with verbs than nouns has been extended to comprehension in progressive nonfluent aphasia (Rhee et al., 2001) and amyotrophic lateral sclerosis dementia (Bak et al., 2001). In both participants with AD and participants with SD, we found a disadvantage for verbs in multiple-choice recognition naming of videos as well, indicating that the word-class effect is present regardless of the material used to present a target. One possible explanation for our findings is that verbs are inherently more difficult than nouns. Analyses of participants performance profiles substantiate the relative deficit for verbs in the vast majority of individuals with AD and SD, although we found some exceptions to this pattern. Some published studies also describe greater difficulty with nouns than verbs in patients with AD (Fung et al., 2001; G. Robinson et al., 1999; Williamson et al., 1998). Daniele and colleagues (Daniele, Giustolisi, & Silveri, 1994; Daniele, Silveri, Giustolisi, & Gainotti, 1993) and Hillis et al. (2004) reported greater difficulty with nouns than verbs in cases of patients with left temporal lobe atrophy and fluent progressive aphasia. Breedin et al. (1994) described marginally worse performance for frequency-matched nouns than verbs on naming-todefinition and synonym judgment tasks in a single case of SD. Observations such as these indicate that verbs are not necessarily more difficult than nouns, although we cannot rule this out definitively for the set of stimuli used in the present study because of the ceiling effect for nouns in control participants. Experimental studies have examined a nonlinguistic basis for relative verb difficulty in several ways. For example, performance on tasks such as naming with verbs and word picture matching with verbs does not correlate with measures of executive resources (K. M. Robinson et al., 1996; White-Devine et al., 1996). Using a priming technique, other researchers have confirmed greater difficulty with verbs compared with nouns in patients with AD (Bushell & Martin, 1997), suggesting that task-related resources cannot explain the deficit with verbs in patients with AD. In a dual-task study of word picture matching with verbs and nouns, performance of a secondary task interfered nonselectively with word picture matching for nouns and verbs in patients with a disorder of social comportment and executive functioning because of FTD (Rhee et al., 2001). In this same study, patients with progressive nonfluent aphasia showed selective difficulty with verbs during concurrent secondary task performance, but such an effect was not seen for patients with SD. In the present study, working memory difficulty is unlikely to explain the verb impairment in patients with AD and SD because the stimuli and response choices remained available in the visual modality. Another possibility may be related to the modality of presentation. Previous work shows an advantage for verb action videos over static line drawings of objects in patients with AD (Fung et al., 2001). However, the use of line drawings to present objects may have disadvantaged these stimuli for participants with AD because the line drawings are not colored or set in a rich, real-world context. In the present study, when videos were used to present both actions and objects, we found worse performance with verbs than nouns. Several additional hypotheses that have been forwarded to explain a disadvantage for verbs focus on the linguistic properties of this word class. In a careful study of verb and noun production during narrative description of the Cookie Theft Scene, Bird et al. (2000) characterized performance in 3 participants with SD longitudinally and in a cross-sectional analysis of 21 protocols in participants with SD. These investigators found greater difficulty with nouns than verbs. To account for the inability to control frequency during production of a semistructured speech sample, Bird et al. assessed the frequency of content words that were spontaneously produced during the narratives. Participants with SD showed a loss of low frequency nouns compared with controls, but the rare occurrence of low frequency verbs in patients with SD allowed them to demonstrate an apparent preservation of verb use. The frequency effect described by Bird et al. is entirely dependent on the corpus of nouns and verbs that the participants decided to produce. This is further complicated by the fact that nouns are often replaced by pronouns without compromising sentence structure, and this substitution may be particularly common in participants with SD who have considerable difficulty retrieving content words. In the present study, we assessed performance for more frequent and less frequent segments of the stimulus set, but we found no specific effect for frequency. Moreover, the impairment with verbs is less likely to be because of a lexical retrieval deficit in the present study because we used a multiple-choice, recognition-naming procedure. In a previous study of patients with AD, we used word picture matching with frequency-matched sets of nouns and verbs (White-Devine et al., 1996) and found greater

7 REVERSED CONCRETENESS EFFECT FOR VERBS 15 difficulty with verbs compared with nouns. These observations raise the possibility that frequency modulates the lexical retrieval process in production measures like those of Bird et al. (2000). Patients with SD are known to have reading difficulty (Noble, Glosser, & Grossman, 2000; Patterson & Hodges, 1992), but this is unlikely to account for the verb deficit in patients with SD because the stimulus materials in the present study also were presented aurally. We hypothesized instead that the deficit for verbs in participants with AD and SD was related more directly to a component of semantic memory. The findings of the present study thus are consistent with the results of K. M. Robinson et al. (1996) and White-Devine et al. (1996), who have proposed that the vulnerability of the neural representation of verbs results in part from a difference in the hierarchical taxonomic organization of verbs compared with nouns. From this perspective, nouns have a well-defined, deeply layered hierarchical structure that is supported by a well-organized, richly interconnected network. Verbs, however, have a shallow and poorly organized semantic hierarchy relative to nouns (Miller & Fellbaum, 1991). According to some investigators, the rich hierarchical structure of nouns allows shared semantic features to compensate for overlapping and semantically related concepts in which feature knowledge is degraded in a neurodegenerative condition, such as AD or SD (Gonnerman et al., 1997; Rogers et al., 2004). Rogers et al. (2004) emphasized the importance of the frequency, familiarity, and semantic density of words in patients with SD. They argued that the density of the semantic neighborhood is an important factor in defining performance on semantic tasks. Reduced density in verb semantic networks may contribute to poor performance with verbs. Because verbs are less well organized hierarchically and therefore have limited neighborhood density (Miller & Fellbaum, 1991), the semantic network has greater difficulty identifying shared features with overlapping concepts and therefore can compensate for degraded features less efficiently. This would allow a deficit for verbs to be more prominent or to emerge relatively earlier in the course of disease than a deficit for nouns. Regardless of the basis for the significant impairment with verbs in participants with SD and participants with AD, we found a qualitatively important difference in verb comprehension across these two groups of participants. In particular, the concreteness of the stimuli significantly influenced performance. We found prominent reversal of the concreteness effect for verbs in participants with SD but not in participants with AD. In controls, a concreteness effect was frequently observed (Paivio, 1991). The concreteness effect has been investigated occasionally in patients with AD for nouns, and in rare case studies in patients with SD, but this effect has not been examined previously for verbs. Previous studies thus have shown significant difficulty with abstract relative to concrete stimuli in patients with AD (Kroll & Merves, 1986; Rissenberg & Glanzer, 1987). In the present study, we also found that participants with AD had worse performance with abstract nouns than concrete nouns as a group, although individual participant analyses did not confirm the group effect. We found marginally greater difficulty with abstract nouns than concrete nouns in participants with SD as well, although 3 of the participants with SD in this study showed reversal of the concreteness effect for nouns. Reversal of the concreteness effect for nouns has been described occasionally in patients with SD (Breedin et al., 1994; Cipolotti & Warrington, 1995; Sirigu et al., 1991; Warrington, 1975, 1981; Warrington & Shallice, 1984). Because SD is associated with disease in the left inferior temporal cortex (Galton et al., 2001; Gorno-Tempini et al., 2004; Grossman et al., 2004), relative difficulty with concrete nouns in some participants with SD may be due in part to the degradation of visual perceptual knowledge underlying objects that is stored in visual association cortex in this brain region. This temporal area also correlates with performance on semantic memory tasks in patients with SD (Mummery, Patterson, Price, & Hodges, 2000; Williams, Nestor, & Hodges, 2005). There would be less of an impact on abstract nouns, from this perspective, because abstract nouns depend much less on visual perceptual knowledge than concrete nouns. Such a visually based account does not necessarily detract from observations suggesting that semantic memory is modality neutral in nature (Bozeat, Lambon Ralph, Patterson, Garrard, & Hodges, 2000; Bozeat, Lambon Ralph, Patterson, & Hodges, 2002), because a rich network of multimodal knowledge may become increasingly vulnerable as the visual-based core of concrete object knowledge is degraded. Reversal of the concreteness effect may be less evident in patients with AD because visual perceptual feature knowledge appears to be less vulnerable to disease in these patients. Performance on semantic memory measures in patients with AD thus correlates with the posterolateral temporal-inferior parietal region (Desgranges et al., 1998; Grossman, Koenig, Glosser, et al., 2003; Grossman et al., 1997; Harasty et al., 1999; Penniello et al., 1995; Prvulovic et al., 2002), an area that is not considered to be visual association cortex. Many fmri studies have recruited visual association cortex during probes of concrete objects (Joseph, 2001), although additional work is needed to specify the precise relationship between these activations and the role that this area may play in the storage of visual feature knowledge associated with objects represented in semantic memory (Kellenbach, Brett, & Patterson, 2001; Lee et al., 2002; Noppeney & Price, 2002). Although we found occasional patients with SD showing a reversal of the concreteness effect for nouns, worse performance with concrete verbs rather than abstract verbs appeared to be more robust in patients with SD. As with nouns, reversal of the concreteness effect may be present for verbs in patients with SD because of relatively degraded visual perceptual knowledge that plays a larger role in concrete words than abstract words. Functional neuroimaging studies of motion verbs in healthy young adults have recruited left posterior-inferior temporal cortex (Grossman et al., 2002; Perani et al., 1999), and these investigators have speculated that temporal cortical activation is related to the processing of visual perceptual features that contribute to motion verb meaning. Verbs of cognition, by comparison, did not recruit this area. Because SD is associated with disease compromising association cortex in the left inferior temporal lobe (Galton et al., 2001; Gorno-Tempini et al., 2004; Grossman et al., 2004), this may preferentially interfere with visual perceptual knowledge that contributes to motion verbs. In this context, we may have observed a reversal of the concreteness effect more robustly for verbs than nouns because of differences in the structure of verb and noun semantic categories (Miller & Fellbaum, 1991). Because nouns are in a well-organized semantic network, it is possible for preserved knowledge of neighboring nouns to compensate for degraded knowledge that is shared with the target (Gonnerman et al., 1997;

8 16 YI, MOORE, AND GROSSMAN Rogers et al., 2004). However, this kind of compensatory mechanism may be less robust for verbs because of the poorly organized structure of the verb semantic network. We may not have observed the same difficulty with motion verbs in patients with AD because the distribution of their disease in more dorsal portions of lateral temporal cortex is thought to be less related to visual perceptual feature knowledge (Desgranges et al., 1998; Grossman, Koening, DeVita, et al., 2003; Grossman et al., 1997; Harasty et al., 1999; Penniello et al., 1995; Prvulovic et al., 2002). We conclude that reversal of the concreteness effect was seen for verbs in patients with SD because of a combination of two effects. First, the verb semantic network is poorly organized relative to the noun semantic network. Degraded feature knowledge for nouns can be compensated for in part because of the well-organized noun semantic network in which features can be shared across overlapping concepts. However, the poorly organized semantic network for verbs limits similar feature sharing for verbs, rendering verbs more vulnerable to progressive neurodegenerative disease than nouns. Second, visual feature knowledge is degraded in patients with SD because of the anatomic distribution of their disease in visual association cortex. This causes relatively greater difficulty for motion verbs than cognition verbs. Reversal of the concreteness effect may be less evident for nouns because the well-structured noun semantic network supports sharing of features with overlapping concepts, but the poorly organized verb semantic network does not support such compensation. References Bak, T. H., O Donovan, D. G., Xuereb, J. H., Boniface, S., & Hodges, J. R. (2001). Selective impairment of verb processing associated with pathological changes in Brodmann Areas 44 and 4 in the motor neurone disease dementia aphasia syndrome. Brain, 124, Bird, H., Lambon-Ralph, M. A., Patterson, K., & Hodges, J. R. (2000). The rise and fall of frequency and imageability: Noun and verb production in semantic dementia. Brain and Language, 73, Bozeat, S., Lambon Ralph, M. A., Patterson, K. E., Garrard, P., & Hodges, J. R. (2000). Non-verbal semantic impairment in semantic dementia. Neuropsychologia, 38, Bozeat, S., Lambon Ralph, M. A., Patterson, K. E., & Hodges, J. R. (2002). When objects lose their meaning: What happens to their use? Cognitive, Affective, and Behavioral Neuroscience, 2, Breedin, S., Saffran, E. M., & Coslett, H. (1994). Reversal of the concreteness effect in a patient with semantic dementia. Cognitive Neuropsychology, 11, Bushell, C. M., & Martin, A. (1997). Automatic semantic priming of nouns and verbs in patients with AD. Neuropsychologia, 35, Cappa, S. F., Binetti, G., Pezzini, A., Padovani, A., Rozzini, L., & Trabucchi, M. (1998). Object and action naming in Alzheimer s disease and frontotemporal dementia. Neurology, 50, Caramazza, A. (2000). The organization of conceptual knowledge in the brain. In M. Gazzaniga (Ed.), The new cognitive neurosciences (pp ). Cambridge, MA: MIT Press. Caramazza, A., & Shelton, J. R. (1998). Domain-specific knowledge systems in the brain: The animate inanimate distinction. Journal of Cognitive Neuroscience, 10, Cipolotti, L., & Warrington, E. K. (1995). Semantic memory and reading abilities: A case report. Journal of the International Neuropsychological Society, 1, Daniele, A., Giustolisi, L., & Silveri, M. C. (1994). Evidence for a possible neuroanatomical basis for lexical processing of nouns and verbs. Neuropsychologia, 32, Daniele, A., Silveri, M. C., Giustolisi, L., & Gainotti, C. (1993). Categoryspecific deficits for grammatical classes of words: Evidence for possible anatomical correlates. Italian Journal of Neurological Sciences, 14, Desgranges, B., Baron, J.-C., de la Sayette, V., Petit-Taboue, M. C., Benali, K., Landeau, B., et al. (1998). The neural substrates of memory systems impairment in Alzheimer s disease: A PET study of resting brain glucose utilization. Brain, 121, Farah, M. J., & McClelland, J. L. (1991). A computational model of semantic memory impairment: Modality specificity and emergent category specificity. Journal of Experimental Psychology: General, 120, Folstein, M. F., Folstein, S. E., & McHugh, P. R. (1975). Mini-mental state: A practical method for grading the state of patients for the clinician. Journal of Psychiatric Research, 12, Francis, W. N., & Kucera, H. (1982). The frequency analysis of English usage. Boston: Houghton-Mifflin. Fung, T. D., Chertkow, H., Murtha, S., Whatmough, C., Pelouquin, L., Whitehead, V., & Templeman, F. D. (2001). The spectrum of category effects in object and action knowledge in dementia of the Alzheimer s type. Neuropsychology, 15, Galton, C. J., Patterson, K., Graham, K., Lambon-Ralph, M. A., Williams, G., Antoun, N., et al. (2001). Differing patterns of temporal atrophy in Alzheimer s disease and semantic dementia. Neurology, 57, Garrard, P., Lambon-Ralph, M. A., Watson, P., Powis, J., Patterson, K., & Hodges, J. R. (2001). Longitudinal profiles of semantic impairment for living and nonliving concepts in dementia of Alzheimer s type. Journal of Cognitive Neuroscience, 13, Garrard, P., Patterson, K., Watson, P., & Hodges, J. (1998). Category specific semantic loss in dementia of Alzheimer s type. Brain, 121, Gonnerman, L. M., Andersen, E. S., Devlin, J. T., Kempler, D., & Seidenberg, M. S. (1997). Double dissociation of semantic categories in Alzheimer s disease. Brain and Language, 57, Goodglass, H., Kaplan, H., & Weintraub, S. (1983). The Boston Naming Test. Philadelphia: Lea & Febinger. Gorno-Tempini, M. L., Dronkers, N. F., Rankin, K. P., Ogar, J. M., Phengrasamy, L., Rosen, H. J., et al. (2004). Cognition and anatomy in three variants of primary progressive aphasia. Annals of Neurology, 55, Grossman, M., & Ash, S. (2004). Primary progressive aphasia: A review. Neurocase, 10, Grossman, M., Koenig, P., DeVita, C., Glosser, G., Alsop, D., Detre, J., & Gee, J. (2002). Neural representation of verb meaning: An fmri study. Human Brain Mapping, 15, Grossman, M., Koenig, P., DeVita, C., Glosser, G., Moore, P., Gee, J., et al. (2003). Neural basis for verb processing in Alzheimer s disease: An fmri study. Neuropsychology, 17, Grossman, M., Koenig, P. L., Glosser, G., DeVita, C., Moore, P., Rhee, J., et al. (2003). Neural basis for semantic memory difficulty in Alzheimer s disease: An fmri study. Brain, 126, Grossman, M., McMillan, C., Moore, P., Ding, L., Glosser, G., Work, M., & Gee, J. C. (2004). What s in a name: Voxel-based morphometric analyses of MRI and naming difficulty in Alzheimer s disease, frontotemporal dementia, and corticobasal degeneration. Brain, 127, Grossman, M., Moore, P., Koeniig, P. L., Antani, S., McCawley, G., Cross, K., & Kwok, S. (2006). Semantic knowledge and categorization difficulty: A longitudinal study of Alzheimer s disease and frontotemporal dementia. Manuscript submitted for publication. Grossman, M., White-Devine, T., Payer, F., Onishi, K., D Esposito, M., Robinson, K. M., & Alavi, A. (1997). Constraints on the cerebral basis

9 REVERSED CONCRETENESS EFFECT FOR VERBS 17 for semantic processing from neuroimaging studies of Alzheimer s disease. Journal of Neurology, Neurosurgery, and Psychiatry, 63, Harasty, J. A., Halliday, G. M., Kril, J. J., & Code, C. (1999). Specific temporoparietal gyral atrophy reflects the pattern of language dissolution in Alzheimer s disease. Brain, 122, Hillis, A. E., Sangjin, O., & Ken, L. (2004). Deterioration in naming nouns versus verbs in PPA. Annals of Neurology, 55, Hillis, A. E., Tuffiash, E., & Caramazza, A. (2002). Modality-specific deterioration in naming verbs in nonfluent primary progressive aphasia. Journal of Cognitive Neuroscience, 14, Joseph, J. E. (2001). Functional neuroimaging studies of category specificity in object recognition: A critical review and meta-analysis. Cognitive, Affective, and Behavioral Neuroscience, 1, Kable, J. W., Lease-Spellmeyer, J., & Chatterjee, A. (2002). Neural substrates of action event knowledge. Journal of Cognitive Neuroscience, 14, Kellenbach, M. L., Brett, M., & Patterson, K. (2001). Large, colorful, and noisy? Attribute- and modality-specific activations during retrieval of perceptual attribute knowledge. Cognitive, Affective, and Behavioral Neuroscience, 1, Kounios, J., & Holcomb, P. J. (1994). Concreteness effect in semantic processing: ERP evidence supporting dual-coding theory. Journal of Experimental Psychology: Learning, Memory, and Cognition, 20, Kroll, J. F., & Merves, J. S. (1986). Lexical access for concrete and abstract words. Journal of Experimental Psychology: Learning, Memory, and Cognition, 12, Lambon-Ralph, M. A., Patterson, K., Garrad, P., & Hodges, J. R. (2003). Semantic dementia with category specificity: A comparative case-series study. Cognitive Neuropsychology, 20, Lee, A. C. H., Graham, K. S., Simons, J. S., Hodges, J. R., Owen, A. M., & Patterson, K. E. (2002). Regional brain activations differ for semantic features but not categories. NeuroReport, 13, Lu, L. H., Crosson, B., Nadeau, S. E., Heilman, K. M., Gonzalez-Rothi, L. J., Baymer, A., et al. (2002). Category specific naming deficits for object naming and actions: Semantic attribute and grammatical role hypotheses. Neuropsychologia, 40, McKhann, G., Drachman, D., Folstein, M., Katzman, R., Price, D., & Stadian, E. M. (1984). Clinical diagnosis of Alzheimer s disease: Report on the NINCDS ADRDA work group under the auspices of the Department of Health and Human Services Task Force on Alzheimer s Disease. Neurology, 34, Merriam-Webster s international dictionary (3rd ed.). (1986). Springfield, MA: Merriam-Webster. Miller, G. A., & Fellbaum, C. (1991). Semantic networks of English. Cognition, 41, Moss, H. E., & Tyler, L. K. (2000). A progressive category-specific semantic deficit for non-living things. Neuropsychologia, 38, Mummery, C. J., Patterson, K. E., Price, C. J., & Hodges, J. R. (2000). A voxel-based morphometry study of semantic dementia: Relationship between temporal lobe atrophy and semantic memory. Annals of Neurology, 47, Neary, D., Snowden, J. S., Gustafson, L., Passant, U., Stuss, D., Black, S., et al. (1998). Frontotemporal lobar degeneration: A consensus on clinical diagnostic criteria. Neurology, 51, Noble, K., Glosser, G., & Grossman, M. (2000). Reading in dementia. Brain and Language, 74, Noppeney, U., & Price, C. J. (2002). Retrieval of visual, auditory, and abstract semantics. NeuroImage, 15, Paivio, A. (1991). Dual coding theory: Retrospect and current status. Canadian Journal of Psychology, 45, Parris, B., & Weekes, B. (2001). Action naming in dementia. Neurocase, 7, Patterson, K. E., & Hodges, J. R. (1992). Deterioration of word meaning: Implications for reading. Neuropsychologia, 30, Penniello, M. J., Lambert, J., Eustache, F., Petit-Taboue, M. C., Barre, L., Viader, F., et al. (1995). A PET study of the functional neuroanatomy of writing impairment in Alzheimer s disease. The role of the left supramarginal and left angular gyri. Brain, 118, Perani, D., Cappa, S., Schnur, T., Tettamanti, M., Collina, S., Rosa, M., & Fazio, F. (1999). The neural correlates of verb and noun processing: A PET study. Brain, 122, Prvulovic, D., Hubl, D., Sack, A., Melillo, L., Maurer, K., Frolich, L., et al. (2002). Functional imaging of visuospatial processing in Alzheimer s Disease. NeuroImage, 17, Rhee, J., Antiquena, P., & Grossman, M. (2001). Verb comprehension in frontotemporal degeneration: The role of grammatical, semantic, and executive components. Neurocase, 7, Rissenberg, M., & Glanzer, M. (1987). Free recall and word finding ability in normal aging and senile dementia of the Alzheimer s type: The effect of item concreteness. Journal of Gerontology, 42, Robinson, G., Rossor, M. N., & Cipolotti, L. (1999). Selective sparing of verb naming in a case of severe Alzheimer s disease. Cortex, 35, Robinson, K. M., Grossman, M., White-Devine, T., & D Esposito, M. (1996). Category-specific difficulty naming with verbs in Alzheimer s disease. Neurology, 47, Rogers, T., Lambon-Ralph, M. A., Garrard, P., Bozeat, S., McClelland, J. L., Hodges, J. R., & Patterson, K. (2004). Structure and deterioration of semantic memory: A neuropsychological and computational investigation. Psychological Review, 111, Silveri, M. C., Daniele, A., Giustolisi, L., & Gainotti, G. (1991). Dissociation between knowledge of living and nonliving things in dementia of the Alzheimer type. Neurology, 41, Silveri, M. C., & Gainotti, G. (1988). Interaction between vision and language in category-specific semantic impairment. Cognitive Neuropsychology, 5, Sirigu, A., Duhamel, J. R., & Poncet, M. (1991). The role of sensorimotor experience in object recognition. Brain, 114, Tyler, L. K., Russell, R., Fadili, J., & Moss, H. E. (2001). The neural representation of nouns and verbs: PET study. Brain, 124, Warrington, E. K. (1975). The selective impairment of semantic memory. Quarterly Journal of Experimental Psychology, 27, Warrington, E. K. (1981). Concrete word dyslexia. British Journal of Psychology, 72, Warrington, E. K., & Shallice, T. (1984). Category specific semantic impairments. Brain, 107, White-Devine, T., Grossman, M., Robinson, K. M., Onishi, K., & Biassou, N. (1996). Verb confrontation naming and word picture matching in AD. Neuropsychology, 10, Williams, G. B., Nestor, P. J., & Hodges, J. R. (2005). Neural correlates of semantic and behavioural deficits in frontotemporal dementia. Neuro- Image, 24, Williamson, D. J., Adair, J. C., Raymer, A. M., & Heilman, K. M. (1998). Object and action naming in Alzheimer s disease. Cortex, 34, (Appendix follows on next page)

10 18 YI, MOORE, AND GROSSMAN Appendix Target Words and Foils Target word Definition Foils Reverse-role SRTW SRRR Motion verbs flee an action that is a rapid departure away from somewhere pursue walk gather push an action that propels an object away from you pull remove glide throw an action that propels an object through the air catch bounce fetch leave an action where you depart voluntarily expel abandon displace nourish an action meaning to provide with food eat clothe suck receive an action meaning to accept something give grab render follow an action meaning to go after lead surround accompany serve an action meaning to place food in front of someone pay pass equip donate an action meaning to contribute to someone obtain foster redeem sell an action where you offer goods in exchange of money buy convey steal borrow an action where you obtain something on loan lend snatch distribute welcome an action where you receive someone call hug invade banish an action where you force someone to leave depart evacuate fly frighten an action where you make someone afraid cringe tickle clasp kill an action where you cause someone s life to end die exterminate stumble poke an action where you jab at someone with your finger flinch shove shrink lean an action where you rest against someone hold hang bear avoid an action where you try to keep away from someone hit crawl knock ride an action where you are conveyed by someone carry climb drag attract an action where you try to get someone s attention perceive tempt gape Cognition verbs listen a behavior where you make an effort to hear something tell look shout show a behavior where you demonstrate visually watch quote investigate teach a behavior where you give instructions to someone learn preach memorize handle a behavior where you inspect an object with your hands feel slap explore respond a behavior where you reply to a question ask hiss repeat commend a behavior where you show approval and praise thank cherish defer scold a behavior where you criticize someone harshly apologize prescribe mutter persuade a behavior where you win an argument by reasoning believe warn reflect speak a behavior where you say something hear pose observe write a behavior where you form letters with a pen read draw consider explain a behavior where you make something comprehensible to someone understand argue know reject a behavior where you refuse a request propose kneel erase treat a behavior where you buy someone a drink or dinner accept exalt appreciate concede a behavior where you give in to an argument acquire offend win bore a behavior where you make someone weary with dullness ignore babble forget refuse a behavior where you decline to do something offer turn hire accuse a behavior where you blame someone for making an error confess find instruct inherit a behavior where you receive goods from an ancestor bestow earn provide seduce a behavior where you persuade someone to have sex submit talk obey expose a behavior where you display something for viewing recognize examine discover Target word Definition OMTW SRTW SROMTW Concrete nouns liquor a liquid that makes you feel drunk soda heroin milk ankle the joint between your leg and foot wrist shoe sleeve knee the joint between your upper and lower leg elbow trouser shirt cane an object that is a short stick used to aid walking wheelchair pole rocker bullet a round missile used for firing a pistol marble rock cotton-ball costume an object that is clothing you wear at special events like Halloween nude gown underwear dirt a substance that is mud or waste matter water grease shampoo bubble an object that is fragile and balloon shaped ball foam circle ink a substance that is a liquid used for writing eraser pencil droplet piano an object that makes music by hitting keys song trumpet melody boy a young man girl fireman actress snow a substance that comes down in flakes from the sky rain sleigh umbrella bench a piece of furniture used to sit on coffin drawer cemetery glove an object that is clothing to keep your hands warm sock veil hosiery bracelet an object that is jewelry worn around your wrist handcuff necklace jail king a male monarch beggar palace rag candle an object that is a stick of wax lit with a match light bulb cake lamp

11 REVERSED CONCRETENESS EFFECT FOR VERBS 19 Appendix (continued) Target word Definition Foils OMTW SRTW SROMTW injury a wound stitch whiplash needle cat a furry pet that purrs and meows dog tiger wolf eagle a large bird of prey, which is the symbol of America chick mountain egg Abstract nouns affection a state where you have a fond feeling toward another repulsion passion vigor agreement a state where you share an opinion conflict gratitude battle amazement a state of extreme surprise disappointment terror grief anger a state of extreme displeasure with someone gaiety aggression happiness suspicion a state of mistrusting someone confidence hysteria honesty bereavement a state when you feel deeply grieved by the death of someone fun pain celebration blessing a state when someone is saying a prayer over you curse advice misfortune boredom a state of feeling tired with dullness enthusiasm laziness excitement bravery a state of being courageous cowardice foolishness anxiety death a state of ending life birth disease anxiety drunkenness a state of being intoxicated by alcohol soberness unconsciousness reason direction a state meaning the way in which you have to go confusion explanation chaos distinction a state of being different unity distance equality effort a state where you try hard to do something apathy application coma emancipation a state where you get your freedom tradition responsibility law vision a state where you see blindness image ignorance fatigue a state of being very tired energy peace vitality humor a state of being funny dullness pleasure impatience marriage a state of being husband and wife separation fidelity jealousy panic a state of sudden overpowering terror calmness madness relaxation Note. Concern was expressed by a reviewer that five of these stimuli may not be categorized unambiguously (borrowing, frightening, attracting, handling, and writing). We reanalyzed patient performance without these five stimuli. There was no change in the statistical results. We continued to find a Group Word Category (noun, verb) Semantic Feature (concrete, abstract) interaction effect, F(1, 38) 4.34, p.05. Using scores prorated to a total possible correct of 20, we continued to find greater difficulty for motion verbs than cognition verbs in semantic dementia (motion 12.84; cognition 14.72), t(11) 2.36, p.05, but not in Alzheimer s disease (motion 14.97; cognition 14.94), as well as greater difficulty with verbs than nouns in semantic dementia, t(11) 4.50, p.001, and Alzheimer s disease, t(27) 7.99, p.001. SRTW semantically related to target word; SRRR semantically related to reverse role; OMTW opposite meaning to target word; SROMTW semantically related to opposite meaning to target word. Received February 9, 2006 Revision received June 5, 2006 Accepted August 2, 2006