Brain and Language 100 (2007) 79–94 www.elsevier.com/locate/b&l
Action and object processing in aphasia: From nouns and verbs to the eVect of manipulability A. Arévalo a,b,¤, D. Perani e,f, S.F. Cappa e,f, A. Butler c, E. Bates a,d, N. Dronkers g,h a
Center for Research in Language, University of California, San Diego, CA, USA SDSU/UCSD Joint Doctoral Program in Language and Communicative Disorders, San Diego, CA, USA c Psychology Department, Washington University, St. Louis, MO, USA d Department of Cognitive Science, University of California, San Diego, CA, USA e Psychology/Neuroscience, Universitá Vita Salute San RaVaele, Milan, Italy f IBFM, CNR, Segrate Milan, Italy g Center for Aphasia and Related Disorders/Audiology and Speech Pathology, VA Northern California Health Care System, Martinez, CA, USA h University of California, Davis, CA, USA b
Accepted 1 June 2006 Available online 1 September 2006
Abstract The processing of words and pictures representing actions and objects was tested in 21 aphasic patients and 20 healthy controls across three word production tasks: picture-naming (PN), single word reading (WR) and word repetition (WRP). Analysis 1 targeted task and lexical category (noun-verb), revealing worse performance on PN and verb items for both patients and control participants. For Analysis 2 we used data collected in a concurrent gesture norming study to re-categorize the noun–verb items along hand imagery parameters (i.e., objects that can/cannot be manipulated and actions which do/do not involve Wne hand movements). Here, patients displayed relative diYculty with the ‘manipulable’ items, while controls displayed the opposite pattern. Therefore, whereas the noun–verb distinction resulted simply in lower verb accuracy across groups, the ‘manipulability’ distinction revealed a ‘double-dissociation’ between patients and control participants. These results carry implications for theories of embodiment, lexico-semantic dissociations, and the organization of meaning in the brain. © 2006 Elsevier Inc. All rights reserved. Keywords: Aphasia; Noun–verb processing; Picture-naming; Word-reading; Word-repetition; Manipulability; Embodiment; Double-dissociation
1. Introduction DiVerential impairments of word processing according to semantic and grammatical categories have been frequently reported in brain-injured patients. Probably the best known and most often studied case is the noun-verb dissociation. Nouns and verbs have been
*
Corresponding author. Fax: +1 858 822 5097/+39 02 21717558. E-mail address:
[email protected] (A. Arévalo).
0093-934X/$ - see front matter © 2006 Elsevier Inc. All rights reserved. doi:10.1016/j.bandl.2006.06.012
shown to dissociate across diVerent languages (Bates, Chen, Tzeng, Li, & Opie, 1991), across various tasks targeting diVerent input and output modalities (Caramazza & Hillis, 1991; Székely et al., 2005), as well as across several populations, e.g., in patients with Alzheimer’s Disease (AD), fronto-temporal dementia (FTD), corticobasal dementia (CBD), and patients with aphasia, among others (for studies looking at noun–verb diVerences in AD and FTD see e.g., Cappa et al., 1998b and Cotelli et al., in press). In aphasia it has been reported that whereas Broca’s patients tend to display diYculty with verbs while retaining their ability to process nouns, Wernicke’s and anomic patients can display the reverse pattern of
80
A. Arévalo et al. / Brain and Language 100 (2007) 79–94
impairment (Chen & Bates, 1998; Daniele, Giustolisi, Silveri, Colosimo, & Gainotti, 1994; Zingeser & Sloan Berndt, 1990).1 In addition to the noun–verb distinction, several other sets of lexico-semantic items have been suggested to dissociate behaviorally and in the brain, e.g., living vs. non-living items (Warrington & McCarthy, 1987; Warrington & Shallice, 1984), yet results for these categories remain controversial (for a review, see Funnell, 2002). Several accounts point to the importance of item features, i.e., items are organized not by word class, but by their semantic and physical features and the correlations between these (Allport, 1985; Martin & Chao, 2001). Other models argue for the combination of feature types, correlated features and distinguishing features together to determine dissociations (Cree & McRae, 2003). Finally, in more recent research, some have reinterpreted many of these dissociations as being a consequence of other basic sensorimotor dimensions, such as the distinction between manipulable vs. non-manipulable items (Gerlach, Law, & Paulson, 2002; Kellenbach, Brett, & Patterson, 2003; Saccuman, Perani, Bates, Danna, & Cappa, 2003). For example, tools are by deWnition manipulable, while buildings or vehicles typically are not. This diVerence in manipulability may therefore play a crucial role in dissociations seen in performance. This last hypothesis is connected to theories of sensorimotor meaning organization, stemming primarily from research on embodiment (e.g., MacWhinney, 1999). Researchers in this area propose that the network of areas on our sensory and motor strips corresponding to diVerent body parts are activated not only when we produce actions with these body parts, but also when we observe others producing them or observe stimuli which elicit such body partrelated imagery (e.g., Buccino, Binkofski, & Riggio, 2004; Rizzolatti, Fadiga, Gallese, & Fogassi, 1996; Rizzolatti et al., 1996). If true, then by deWnition, the semantic and sensorimotor properties behind a given set of stimuli should play a signiWcant role in the way knowledge is organized in the brain, and thus also in the way language may break down following brain injury. Two recent studies (Hauk & Pulvermüller, 2004; Hauk, Johnsrude, & Pulvermüller, 2004) used ERP and event-related fMRI to test participants on a passive reading task of words related to diVerent body parts. The authors identiWed distinct face-, arm- and leg-area activations for the words related to these diVerent body parts, supporting their hypothesis that the pattern of cortical activation elicited by an action word reXects the cortical representation of the action to which the word refers. Yet another study by the same group (Pulvermüller, Hauk, Nikulin, & Ilmoniemi, 2005) used TMS to directly stimulate the areas in the left-hemisphere that pro1 Classic teaching is not reliably replicated, however. One problem with many study designs is the way in which “good” vs. “bad” performance is deWned without providing clear, statistically-determined evaluations. Many studies also suVer from small sample sizes (indeed, the majority are single case reports). Therefore, predicting performance proWles from conventional (or classic) taxonomies remains largely unreliable.
cess leg, arm and hand actions. Through a lexical decision task, the authors found signiWcant correlations between the response speed to words from the diVerent body part categories and the TMS stimulation of their associated cortical areas in the left hemisphere. Finally, in an fMRI study by Saccuman et al. (2003) using the picture stimuli presented in the current experiment, the authors divided the noun and verb stimuli into ‘manipulable’ and ‘non-manipulable’ groups using subjective categorization criteria. Results suggested that diVerences in brain activation were more robust when items were compared along the manipulability grid than along the lexical noun–verb distinction. This paper presents a behavioral production2 study conducted with healthy and aphasic participants that was analyzed along two diVerent parameters: nouns vs. verbs and ‘manipulable’ vs. ‘non-manipulable’ items. Patients were administered three diVerent production tasks (picture-naming—PN, word-reading—WR, and word repetition paradigm—WRP) which were originally designed along the noun–verb dimension. For Analysis 1, we evaluated participants’ performance on nouns and verbs across the three tasks and looked for possible diVerences in patient and control performances. Analysis 2, on the other hand, probed for any manipulability-based dissociations. To render our ‘manipulability’ categorization as objective as possible, we created a classiWcation scheme using data collected in a concurrent project on gesture norming (see Section 2). 2. Methods 2.1. Participants A total of 41 individuals participated in the study, with a breakdown as follows: 21 individuals with aphasia, 10 agematched control participants, and 10 college-aged control participants. The patient sample contained 10 Anomic, 6 Broca’s and 5 Wernicke’s patients, as classiWed by the Western Aphasia Battery (Kertesz, 1979). They were 4 females and 17 males with a mean age of 64 and an average of 15 years of education (see Table 1 for patient information). The older control group included 6 females and 4 males with a mean age of 64 and a mean education level of 14 years. Finally, the college-aged control group included 6 males and 4 females with a mean age of 19 and an average of 13 years of education.3
2 This is considered a “production” study because participants were required to produce spoken answers as opposed to conWrm their comprehension through other means (e.g., a button press or a “yes/no” response paradigm). 3 We norm all of our data with young, healthy control participants because we consider this important to have as a reference for whichever other population is tested; the young, healthy data represent ‘standard’ or ‘optimal’ performance. Although we already had normative data collected from young controls on these speciWc picture and word items, no young controls had been tested on this particular design. College and older control data diVered only slightly from each other in this task, yet all analyses presented here which compare control and patient groups included the older age-matched controls only.
A. Arévalo et al. / Brain and Language 100 (2007) 79–94
2.2. Stimuli
Table 1 Patient information Patient ID Gender AphType Site Age TPO Edu Motor MEVect 321105 321084 311106 321082 321107 120716 321108 311074 110988 121010 321085 120778 321109 120979 120892 121014 311110 120951 120822 120970 310940
M M F M M M M M F M M M M M M M F M M M F
81
Anomic Anomic Anomic Anomic Anomic Anomic Anomic Anomic Anomic Anomic Broca Broca Broca Broca Broca Broca Wernicke Wernicke Wernicke Wernicke Wernicke
S S S S S M S S M M S M S M M M S M M M S
81 67 77 66 57 57 75 64 77 56 51 59 65 48 33 59 73 67 79 76 51
145 89 133 83 104 90 65 272 35 13 45 175 60 44 48 13 66 56 93 29 98
14 14 16 18 14 16 14 14 12 15 12 14 19 16 13 16 16 20 12 16 16
Y N N N Y Y N Y Y N Y Y Y Y Y N N N N N N
Y N N Y Y Y Y N Y Y Y N Y Y Y N Y Y Y Y Y
Information is listed by Patient ID code (generated by our laboratory), Gender (M, male; F, female), Aph(asia) Type (or classiWcation), (testing) Site (S, San Diego, CA; M, Martinez, CA), Age (at time of testing), TPO: time post onset at time of testing (in months), Edu: years of education, Motor: presence of signiWcant motor deWcits (i.e., noticeable weakness, hemiparesis or paralysis; Y, yes, N, no; Y cases always involved some weakness of the hand), and M(anip) eVect: performance on manipulable and non-manipulable items (in PN, WRP or both) diVered by at least 5 percentage points (non-manipulable advantage); Y, yes, N, no.
All participants had normal, or corrected to normal, vision, and were tested for hearing with a standard questionnaire and/or with an audiometer. They were all righthanded, native speakers of English (with no signiWcant exposure to another language prior to the age of 12). All control participants were neurologically intact and did not take any medications. Patients underwent a standard clinical neurological evaluation conducted by a board-certiWed behavioral neurologist, which ruled out the presence of dementia. Only patients with a single, identiWable infarct conWned to the left hemisphere were included (as assessed by the neurologist from each patient’s MRI and/or CT scan). All patients included in the analyses were able to complete at least 25% of the task (at least 10 out of 40 items per session). Two additional Wernicke’s patients were tested but were not able to complete more than 25% of the task, and were therefore excluded from the analyses presented here. The college-aged participants were recruited from the UCSD campus, and were given course credit for their participation. The older control participants were recruited from the greater San Diego and San Francisco Bay Area communities, and were paid for their participation. All other individuals (aphasic patients) were recruited from the Veterans’ Administration Medical Centers, San Diego and Martinez, CA, and were also paid for their participation.
Our stimuli include 120 black and white 2-D line drawings representing actions and objects, as well as their written and aurally-presented counterparts. These were drawn from a larger corpus of 795 stimuli (the Center for Research in Language-International Picture Naming Project, CRL-IPNP, Bates et al., 2000). Items were presented to each participant in pre-randomized orders across the three production tasks. These were 60 pictures representing actions and 60 representing objects, with half (30) of the items in each category being ‘easy’ and half ‘diYcult’.4 We called this task the Production Mini-Battery (for more details, see Arévalo, Moineau, Saygin, Ludy, & Bates, 2005b). It is important to note that several investigators have attempted to match actions and objects on a number of parameters. Székely and Bates (2000), for example, have found that matching actions and objects on frequency, age of acquisition or picture complexity results in a mismatch for naming diYculty measures. Likewise, matching for diYculty results in a mismatch on other lexical and pictorial properties. Therefore, for our particular set of items, we decided to match our items on accuracy as much as possible and test for eVects of other factors in post-hoc analyses (see Section 3.3). As mentioned in Footnote 4 above, despite the inherent action-object representation diVerences, our Wnal word category groups did not diVer signiWcantly from each other in terms of accuracy (92% for objects, 90% for actions). 2.3. Procedure As mentioned above, three tasks were chosen for this study: picture-naming (PN), single word reading (WR) and single word repetition (WRP). Testing our stimuli across three diVerent processing tasks (involving both visual and auditory modalities) allows us to better challenge the putative dissociations we are testing. Namely, strong evidence for a dissociation along grammatical lines would emerge if clear noun–verb diVerences were manifested across all three tasks. On the other hand, dissociations occurring in only one task (or one modality) would constitute weak evidence for such a view. Timed picture-naming (PN) is among the Wrst paradigms ever used to study real-time language processing 4 DiYculty was determined by healthy, college-aged control participants’ RTs on previous norming studies using the same stimuli. We Wrst ‘bracketed’ the set of possible pictures using response accuracy: here, objects needed to have an accuracy score of at least 80% and actions 60% (the lower number for actions reXects the paucity of actions named with high accuracy). The mean accuracy scores for the Wnal samples used in this study, however, did not diVer signiWcantly from each other (92.2% for objects and 89.7% for actions). A subset of these items were then assigned to diYculty bins, where all items with mean RTs of at least two standard deviations above the grand mean classed as ‘diYcult’, and items with RTs at least two standard deviations below the mean classed as ‘easy’.
82
A. Arévalo et al. / Brain and Language 100 (2007) 79–94
(Székely et al., 2005). To name a picture, one must recognize a particular concept from an image, derive its speciWc meaning, and link that meaning to its appropriate label, which in turn must be produced. Various models have been proposed to account for our ability to perform such a task, each with its own set of processing stages (for more details, see Arévalo et al., 2005b). Reading (WR) and repetition (WRP), on the other hand, most likely involve quite diVerent types of processing. WRP varies from the other two tasks in that it is a purely auditory task. Thus, performance on WRP should be signiWcantly inXuenced by word–form properties manifested at the spoken level, such as word length and initial letter sound. WR should similarly be aVected by such word–form properties when these are manifested visually (e.g., word length). Finally, PN, while also visual, is strongly aVected by semantic variables, since this task (arguably more so than the other two) requires one to process the meaning behind the image in order to produce the appropriate response. Other factors considered to be particularly important for PN are ‘goodness of depiction’ of the pictures used, word frequency of their corresponding written or spoken forms and age of acquisition (these last two have also been found to inXuence reading, Bates, Burani, D’Amico, & Barca, 2001). In our posthoc analyses we address this issue by testing which factors most inXuenced each subject’s performance on each modality, thus better revealing the nature of each modality (see Sections 3 and 3.3). For the PN condition in the current study, the picture items were presented one at a time on the computer screen; for the WR condition, each target name5 for the selected pictures was presented as a static word on the computer screen, and for the WRP condition, the same target words were presented aurally via two small speakers attached to the testing computer while the screen remained blank. In addition, aurally-presented lead-in sentences preceded each picture, written word and auditory presentation of the target words. All sentences (e.g., ‘I want to, ƒ, ‘Look at this, ƒ’) were predictive or facilitative of the lexical category of the word they preceded and were acquired from Liu (1996). Each participant was instructed to sit in front of the computer and attend to the pictures, written words and prerecorded spoken stimuli, which appeared one at a time in separate lists of 40 items (each list included stimuli from one condition only). Participants Wrst heard a lead-in sentence and were then required to name the picture, read or repeat the word (depending on the list) that was presented to them. They were asked to respond as accurately as possible into a standing microphone placed in front of them as each stimulus was presented. They were told that some stimuli would represent objects and others actions, and that 5 Target names are the dominant responses elicited by healthy participants for the same picture stimuli on previous noun-only and verb-only PN runs.
these would be in random order. They were also asked to provide their best guess when not sure of the answer. Three blocks of trials were presented. For each block, participants Wrst viewed 40 items in one condition (e.g., PN), then 40 in another (e.g., WR), and the Wnal 40 in the last condition (e.g., WRP). The lists of 40 were rotated for the second and third blocks, respectively, such that across all three blocks, each participant viewed all 120 items three times, with each item appearing once in each of the three conditions. The order of lists and blocks was pre-randomized so that not all participants viewed the same order of trials. The task was presented using PsyScope (Cohen, MacWhinney, Flatt, & Provost, 1993) and was experimentercontrolled (i.e., the experimenter manually advanced each trial). Three practice items were incorporated at the beginning of each list in order to acquaint participants with the task. They were given breaks between lists or changes of condition if they so wished. If no response was given on a given trial, the microphone would not detect a sound, an “X” would appear on the screen above the trial number and the experimenter could then move on to the next trial. 2.4. Scoring Response times were recorded by PsyScope and accuracy was recorded by the experimenter, who manually wrote down all answers during the experiment. Generally, an answer is considered correct if it is the appropriate target word and is accurately detected by the microphone. However, many answers produced by our patients were preceded by non-relevant sounds caused by circumlocutions, false starts or anomias, leading to a high percentage of non-usable RTs. Therefore, we decided to exclude the RT measure and focus on accuracy for the results we report here. For patients, special attention was given to individual answers to allow for future error analyses. Also, some phonological variations of the target answer were considered correct (if similar enough to the target concept). We applied the method used by Martin, Dell, SaVran, and Schwartz (1994): if at most two incorrect phonemes were produced and the answer was identiWable by the experimenter independently of the context, the answer was considered correct. Steering away from the strict scoring method also allows us to consider alternative answers produced by our participants, thus creating a more accurate view of performance abilities and diYculties. 3. Results Analyses were conducted using the JMP software (SAS Institute Inc., 4.0, 1989–2002). We conducted Analysis of Variance tests on the accuracy and RT performances of all Wve groups (and relevant interactions between eVects) as well as individual t-tests between diVerent pairs of groups as well as diVerent task, word category and manipulability
A. Arévalo et al. / Brain and Language 100 (2007) 79–94
83
Table 2 Analysis 1 (noun-verb): signiWcant interactions in accuracy Interaction
F-value
p-Value
G£C G£D G£W C£D C£W G£C£D G£C£W C£W£D
F(1, 8) D 40.96 F(1, 4) D 10.06 F(1, 4) D 2.41 F(1, 2) D 104.50 F(1, 2) D 53.60 F(1, 8) D 4.35 F(1, 8) D 2.78 F(1, 2) D 4.06
p < .0001 p < .0030 p < .0466 p < .0001 p < .0001 p < .0001 p < .0045 p < .0172
G, group, C, condition (or task), D, diYculty, W, word category (Noun/ verb).
eVects within groups (see Sections 3.1 and 3.2). As mentioned in the Scoring section above, we decided to exclude the RT analyses in the Wnal results due to a high percentage of missing data (false starts or failure of the computer to pick up on answers) and to the fact that we believe accuracy information is more signiWcant in terms of portraying true performance (especially in the case of patients).
Fig. 1. Group £ Condition. All groups were signiWcantly less accurate in PN. Performance on WR and WRP did not diVer in control groups and Anomics, yet severe aphasics were worse on WR relative to WRP. These groups also showed greater discrepancy in performance between PN and the other two tasks, and this eVect was strongest for the Wernicke group. Error bars represent standard error of the mean (SEM).
3.1. Analysis 1: Noun–verb As expected, groups diVered greatly from each other on overall task performance: across the three conditions, control groups displayed the highest accuracy (with the college group performing slightly better than the older control participants), followed by the Anomics, then Broca’s and Wnally Wernicke’s patients (F(1, 4) D 516.8056, p < .0001). In addition, there was a main eVect of Condition: all groups were signiWcantly less accurate at PN (F(1, 2) D 500.9318, p < .0001). In the case of both control groups and anomic patients, performance on WR and WRP was virtually the same; severe patients (Broca’s and Wernicke’s), on the other hand, were signiWcantly more accurate at WRP relative to WR. A main eVect of word category was also signiWcant overall (F(1, 1) D 26.7899, p < .0001), with all participants responding signiWcantly more accurately to nouns relative to verbs. Finally, there was a main eVect of DiYculty (F(1, 1) D 199.1002, p < .0001), with items originally classiWed as ‘easy’ yielding comparatively more accurate scores. Table 2 includes a list of signiWcant 2- and 3-way interactions. PN was the most challenging task for control and patient groups alike, yet patients revealed comparatively greater diYculty on it (relative to control groups as well as to their own performance on the other two tasks). This result is illustrated in Fig. 1. In addition, poorer performance on verb processing was only signiWcant in the PN condition (F(1, 4) D 4.1614, p < .0023). In other words, verb pictures were less accurately named than noun pictures, but verb words and sounds did not diVer signiWcantly from written and aurally-presented nouns (WR, F(1, 4) D .8538, p < .491; WRP, F(1, 4) D .2368, p < .9177). Fig. 2 illustrates the noun–verb dissociation across groups in PN.
Fig. 2. PN, Group £ Word Category. In PN, all participants were signiWcantly less accurate at processing verbs, with patient groups displaying greater discrepancy between the categories. F(1, 4) D 4.1614, p < .0023. Error bars represent standard error of the mean (SEM).
Finally, a set of posthoc analyses were conducted to assess whether other relevant variables inXuenced participants’ performance on the three tasks. These were number of syllables and number of letters (most relevant for WR and WRP), initial letter frication, word frequency, AoA (age of acquisition), and visual complexity (relevant mostly for PN). No signiWcant eVects on performance were found for any of these variables. 3.1.1. Noun–verb results summary Our main goal with this analysis was to compare group performance across conditions and test the classic noun–verb double-dissociation previously reported for aphasic patients. Our results were as follows: Wrst, as expected, PN was more diYcult for all participants tested; therefore, accuracy on this task was signiWcantly lower relative to WR and WRP Table 3. However, aphasic groups showed particular diYculty with PN, with their performance dropping signiWcantly on this task relative to the other two, suggesting perhaps a lower breaking point (or diYculty threshold) for patients relative to healthy controls. Finally, WR and WRP seemed of equal
84
A. Arévalo et al. / Brain and Language 100 (2007) 79–94
Table 3 Analysis 1 (noun–verb): main eVects for accuracy by group Participant group
N > V (across all three tasks)
N > V (in PN only)
College controls Older controls Anomic patients Broca’s patients Wernicke’s patients All controls All patients
F(1, 1) D 13.4726, p < .0002
F(1, 1) D 15.6479, p < .0001
F(1, 1) D 8.1483, p < .0043
F(1, 1) D 14.8364, p < .0001
F(1, 1) D 7.6922, p < .0056
F(1, 1) D 17.1901, p < .0001
F(1, 1) D 11.399, p < .0007
F(1, 1) D 32.0104, p < .0001
Missed signiWcance
F(1, 1) D 4.9967, p < .0258
F(1, 1) D 20.2613, p < .0001 F(1,1) D 15.4418, p < .0001
F(1, 1) D 30.1743, p < .0001 F(1, 1) D 97.0317, p < .0001
N > V, better performance on nouns relative to verbs.
Using the analyzed gestures of the participants who responded to these items and who have been completely analyzed thus far (n D 32), we re-categorized the items used in the Mini-Battery experiment. Of 120 noun and verb items, 49 were classiWed as ‘manipulable’ (24 verbs and 25 nouns) and 71 were ‘non-manipulable’ (36 verbs and 35 nouns). In addition, a second set of analyses was conducted in which any items which also elicited signiWcant involvement of other major body parts (i.e., mouth, foot and whole body; e.g., Ehrsson, Geyer, & Naito, 2003) were identiWed and excluded (see Section 3.3). Having originally asked the noun-verb question, we then asked whether any of our participants performed diVerently when naming, reading or repeating ‘manipulable’ vs. ‘non-manipulable’ items. 3.2. Analysis 2: Manipulability
diYculty for controls and Anomics (the least severe of the patient groups), yet both severe groups (Broca’s and Wernicke’s patients) were signiWcantly more accurate at WRP relative to WR. In addition, PN was the only condition in which there was a signiWcant discrepancy in noun vs. verb performance, suggesting this type of word class dissociation may be speciWc to word-retrieval processes. Furthermore, the type of dissociation seen was consistent across control and patient groups: verbs were always named less accurately. Therefore, these results do not support theories arguing that diVerent aphasia classiWcations lead to opposing performance proWles (at least in the case of noun and verb processing). As mentioned above, our next step was to investigate whether another dimension—manipulability—would reveal more robust dissociations. To provide a more objective way of categorizing items into ‘manipulable’ vs. ‘nonmanipulable’ groups, we developed an additional norming measure which taps into a number of diVerent word and gesture variables, one of which is hand involvement. This measure is described below: 3.1.2. Gesture norming (Arévalo, Butler, Perani, Cappa, & Bates, 2005a) In this study, we presented participants with the written form of the target names of the Bates et al. (2000) stimuli and asked them to ‘do the Wrst thing that comes to mind’ when thinking of that word (i.e., a movement, a pose, etc.). Participants were videotaped and their gestures were coded and recorded into the database. Our elaborate coding program was designed along neurological criteria, and allows us to record, for example, whether a movement performed is proximal or distal to the body, whether it is simple or complex, whether it is Wne or gross, and which body parts were involved. An item was considered ‘manipulable’ if the majority (at least 70%) of participants who gestured to it produced obvious Wne-grained movements of the Wngers (especially in a grasping motion). All other items were classiWed as ‘non-manipulable’.
For Analysis 2, we began by investigating participants’ overall accuracy on manipulable (nouns and verbs) vs. non-manipulable (nouns and verbs) items, collapsed across all three conditions (Fig. 3a). While both control groups were signiWcantly more accurate at processing the manipulable items (College, F(1, 1) D 7.0642, p < .0079; Older, F(1, 1) D 13.3653, p < .0003), patients as a group were signiWcantly more accurate at processing the nonmanipulable items (F(1, 1) D 7.8611, p < .0051). When analyzed separately, this eVect reached signiWcance for both severe groups: Broca’s (F(1, 1) D 5.5445, p < .0186) and Wernicke’s patients (F(1, 1) D 6.5022, p < .0109). Anomic patients went in the same direction, but the eVect did not reach signiWcance (F(1, 1) D 0.0178, p < .8938). In addition, the interaction of Group £ Manipulability for Control participants did not reach signiWcance (F(1, 1) D 1.614, p < .2040), indicating that College and Older control performance patterns did not diVer signiWcantly. An interaction of Subgroup (Older controls vs. Patients) £ Manipulability, however, did reach signiWcance (F(1, 1) D 12.687, p < .0004, see Fig. 3b). Fig. 3c and d illustrate the interactions between Older controls vs. Broca’s patients and Older controls vs. Wernicke’s patients, respectively (Table 4). Next, we broke performance down by task to see where groups most strongly displayed this manipulability-based eVect. When PN was analyzed individually, the eVect reached signiWcance for both control groups (better performance on manipulable vs. non-manipulable items: College, F(1, 1) D 7.4705, p < .0064, Older, (F(1, 1 ) D 10.0057, p < .0016), as well as for Wernicke’s patients (better performance on non-manipulable relative to manipulable items: F(1, 1) D 7.783, p < .0054). An interaction of Subgroup (Older controls vs. Patients) £ Manipulability for PN was also signiWcant (F(1, 1) D 6.4924, p < .0109, see Fig. 4a), as well as the interaction of Manipulability in PN for Older controls vs. Wernicke’s patients (F(1, 1) D 19.4863, p < .0001, see Fig. 4b). In the case of WRP, College control performance was at ceiling, while
A. Arévalo et al. / Brain and Language 100 (2007) 79–94
a
b
c
d
85
Fig. 3. (a) All tasks, Group £ Manipulability. Each group’s performance on manipulable vs. non-manipulable items collapsed across word categories and tasks. The interaction of Group £ Manipulability was signiWcant (F(1, 4) D 7.8891, p < .0001; removing the college control group: F(1, 3) D 7.4223, p < .0001), as well as the interaction of Subgroup £ Manipulability, comparing Older controls to Patients analyzed together as a group (F(1, 1) D 12.687, p < .0004). The diVerence in healthy and patient performance revealed a double-dissociation in the processing of manipulable items. Anomic patients (the least severe group) were in line with severe aphasic performance, but the diVerence did not reach signiWcance. Error bars represent standard error of the mean (SEM). (b) All tasks: Subgroup £ Manipulability. Here older control participants were compared to patients taken as a group. The interaction was signiWcant (F(1, 1) D 12.687, p < .0004). Error bars represent standard error of the mean (SEM). (c) All tasks, Manipulability in Older controls vs. Broca’s patients. As in Older controls vs. all patients (3a), Older controls vs. Broca’s patients resulted in a signiWcant interaction (F(1, 1) D 16.5893, p < .0001). (F(1, 1) D 12.687, p < .0004). Error bars represent standard error of the mean (SEM). (d) All tasks, Manipulability in Older controls vs. Wernicke’s patients. This interaction was also signiWcant (F(1, 1) D 19.6222, p < .0001). Error bars represent standard error of the mean (SEM).
Table 4 Analysis 2 (Manipulability): signiWcant interactions Interaction: Subgroup £ Manip
F-value
p-Value
Overall PN Reading WRP
F(1,1) D 14.2863 F(1,1) D 4.9792 F(1,1) D 4.7454 F(1,1) D 6.8102
p < .0002 p < .0257 p < .0295 p < .0091
Manipulability for subgroups (Older controls vs. Patients), overall and in each task separately. These results represent the Conservative analyses, in which all items that involved other body part imagery were excluded.
Older controls displayed a signiWcant manipulability eVect (better repetition of manipulable than non-manipulable items: F(1, 1) D 5.3486, p < .0209). For patients, the eVect (better repetition of non-manipulable items) reached signiWcance when all patients were analyzed together (F(1, 1) D 6.2696, p < .0123) as well as for Broca’s patients when this group was analyzed separately (F(1, 1) D 6.6664, p < .0100). Finally, the interaction of Subgroup (Older controls vs. Patients) £ Manipulability for WRP reached signiWcance (F(1, 1) D 6.4872, p < .0109, see Fig. 5a), as well as Older controls vs. Broca’s patients (F(1, 1) D 14.2869, p < .0002, Fig. 5b). Results for WR went in the same direction as the other two conditions, but no eVects reached signiWcance.
Next, we broke down our analyses one step further to see if diVerences in performance had diVerent eVects on our two original categories: nouns vs. verbs. In the case of control groups, both College and Older groups displayed a signiWcant manipulability eVect in objects (nouns) and not in actions (verbs) across tasks (see Fig. 6a and b; the interaction of word category £ manipulability was not signiWcant, however). In the case of patients, on the other hand, the eVect was signiWcant for actions (verbs) and not for objects (nouns), when all patients were analyzed together as one group (see Fig. 6c). No eVects reached signiWcance when patient groups were analyzed separately either overall or for each condition separately. As in the noun–verb analyses, no signiWcant performance diVerences for manipulability were observed in the reading (WR) task. As reported above, both control groups and anomic patients in this sample displayed high accuracy (equal to WRP performance) in the WR task, while both severe patient groups were more accurate at WRP than WR. It has been argued that one can achieve relatively good reading performance without complex processing or high comprehension (as long as enough phonological memory is available; e.g., Devlin et al., 2002). However, error analyses for the current study suggest that control groups and ‘low-impairment’ aphasic
86
A. Arévalo et al. / Brain and Language 100 (2007) 79–94
a
b
Fig. 4. (a) PN, Manipulability in Older controls vs. Patients as a group. Also in PN, control performance on manipulable vs. non-manipulable items showed the opposite pattern as that of patients when analyzed together as one group (F(1, 1) D 6.4924, p < .0109). Error bars represent standard error of the mean (SEM). (b) PN, Manipulability in Older controls vs. Wernicke’s patients (F(1, 1) D 19.4863, p < .0001). Error bars represent standard error of the mean (SEM).
patients (i.e., Anomics) may employ processing strategies which are not readily available to more impaired individuals. For example, several incorrect answers produced by patients in the WR task were of a semantic rather than phonological nature (e.g., “happy” rather than “smile”). In addition, Broca’s patients have been reported to have particular diYculty with reading. Therefore, patients’ overall diYculty in WR may reXect poorer strategies, which in turn lead to lower accuracy. In sum, the WR task did not reveal any signiWcant dissociations in either of our analyses (Nouns vs. verbs or Manipulability), and whereas a unidirectional noun–verb dissociation was seen for PN, Analysis 2 suggests the presence of a double manipulability-based dissociation in two of the tasks: PN as well as WRP. 3.3. Additional analyses In this work, we chose to focus on one body part (i.e., the hand) which we felt would be signiWcantly involved in the processing of these stimuli according to Embodiment Theory
and which we expected would be signiWcantly represented among our items. However, as mentioned above, other body parts other than the hand are also considered critical in such sensorimotor representations, i.e., the foot (or toes) and the mouth (or tongue) (Ehrsson et al., 2003). Some of our items (marked with an asterisk in Table 5: Object items and Table 6: Action items) elicited signiWcant hand involvement but also elicited the involvement of other key body parts, as seen in the Gesture Norming data. Our original analyses were run on all 120 items, yet we recognized that these ‘multiple body parts’ items could introduce signiWcant confounds. Therefore, we ran a second set of more ‘conservative’ analyses having excluded these items. The remaining item set consisted of 48 ‘manipulable’ items (23 actions and 25 objects) and 40 ‘non-manipulable’ items (14 actions and 26 objects) Fig. 7. For these ‘conservative’ analyses, all previously signiWcant eVects remained signiWcant. As in the previous analyses, patients were still signiWcantly less accurate at processing the manipulable items (with a stronger impact on the actions) and control participants were signiWcantly more accurate at processing the manipulable items. In addi-
A. Arévalo et al. / Brain and Language 100 (2007) 79–94
87
a
b
Fig. 5. (a) WRP, Manipulability in Older controls vs. Patients (F(1, 1) D 6.4872, p < .0109). Error bars represent standard error of the mean (SEM). (b) WRP, Manipulability in Older controls vs. Broca’s patients (F(1, 1) D 14.2869, p < .0002). Error bars represent standard error of the mean (SEM).
tion, one new result emerged: the manipulability eVect became signiWcant for Broca’s patients in the WR condition (F(1, 1) D 6.2217, p < .0129). Finally, we ran posthoc analyses on a set of item features known to inXuence processing performance (as mentioned above for Analysis 1). These have been calculated into our item corpus and include objective visual complexity (of the picture), and word frequency, age of acquisition, number of syllables, number of characters (i.e., word length), and initial letter frication (of the word form of each item). We had controlled for these factors in our original noun-verb item selection, and repeated the analyses this time to consider them within the manipulability dimension. Comparing these variables once again for our new manipulability categories revealed that none of them accounted signiWcantly for any results attained. In addition, we wanted to make sure our items were relatively well-balanced for the diYculty variable (not controlled for originally, given that manipulability was introduced as a variable later in the study). Surprisingly, the diYculty measure turned out to be almost perfectly balanced. In the Conservative Analyses, out of the 40 ‘manipulable’ items, 19 were
‘easy’ and 21 were ‘diYcult’; similarly, for the 48 ‘non-manipulable’ items, 24 were ‘easy’ and 24 ‘diYcult’. The interaction of Manipulability £ DiYculty was not signiWcant (F(1, 1) D 3.3222, p < .0684). 3.4. Probing lesion information Our next thought was to test whether any neurological information on our patients could explain any of our behavioral Wndings to any degree. Most neuroimaging analysis methods (i.e., VLSM, Bates et al., 2003) would require us to have a larger patient sample (at least 25 patients) to reach any reliable lesion-performance correlations. Therefore, we decided to use two diVerent preliminary approaches instead which may predict future Wndings with a larger patient sample. First, we were interested in seeing how many of the patients showing the manipulability eVect6 in PN, WRP or 6 We determined that an individual patient displayed the eVect signiWcantly if their accuracy score on the manipulable items for a given task was at least 5 percentage points lower than their score on the non-manipulable items.
88
A. Arévalo et al. / Brain and Language 100 (2007) 79–94
a
b
c
Fig. 6. (a) College controls, Manipulability £ Word Category. In the case of College controls, the manipulability eVect reached signiWcance in the case of objects (nouns) (F(1, 1) D 8.2546, p < .0041) but not actions (verbs), across tasks. The Manipulability £ Word Category interaction did not reach signiWcance. Error bars represent standard error of the mean (SEM). (b) Older controls, Manipulability £ Word Category. As for college students, Older controls displayed a signiWcant manipulability eVect for objects (nouns) and not for actions (verbs) across tasks (F(1, 1) D 12.3437, p < .0005), yet the Manipulability £ Word Category interaction did not reach signiWcance. Error bars represent standard error of the mean (SEM). (c) Patients, Manipulability £ Word Category. Contrary to the two control groups, patients displayed a stronger manipulability eVect in actions (verbs) relative to objects (nouns) (manip. eVect for verbs, F(1, 1) D 6.2437, p < .0125). The Manipulability £ Word Category interaction did not reach signiWcance.
A. Arévalo et al. / Brain and Language 100 (2007) 79–94
89
Table 5 Object items No.
Picture name
RT-total
RT-target
Ln frequency
VisComplexity
Manip
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
ACCORDION AIRPLANE AX BABY BALLOON BED BELL BICYCLE BREAD¤ (m) CANNON CARROT¤ (m) CHIMNEY CLOCK COMB COW DOOR DRUM FAUCET FISHINGROD FORK¤ (m) GLASSES HANDCUFFS HAT IRONINGBOARD KITE LIZARD MOON ONION ORANGE¤ (m) PACKAGE PANTS PURSE RABBIT RING SANDWICH¤ (m) SAXOPHONE¤ (m) SCREWDRIVER SEAHORSE SEAL SHELL SKELETON SLED SMOKE SNAKE SOCK¤ (f) SQUIRREL SUBMARINE TANK TELEPHONE¤ (m) TELEVISION TENT TURKEY TURTLE UMBRELLA UNICYCLE¤ (f) VASE WAITER WHEELBARROW WORM YOYO
1216 800 1119 751 702 706 703 751 774 1159 806 1169 776 717 1115 719 779 1168 1231 723 766 1139 692 1110 796 1229 804 1115 1129 1088 779 780 742 785 775 1103 1179 1157 1221 1129 817 1198 1212 775 712 1225 1144 1181 761 799 744 1159 734 738 1173 1168 1161 1226 1106 1155
1179 778 1085 729 702 706 703 731 773 1159 806 1169 772 717 1079 719 766 1130 1213 723 758 1113 684 1105 796 1155 804 1100 1098 1102 757 772 746 785 775 1061 1179 1132 1115 1101 817 1188 1221 775 712 1234 1145 1155 752 786 744 1160 734 738 1179 1171 1156 1207 1110 1141
0.69 1.95 2.3 5.56 1.95 5.14 3.33 1.79 4.32 1.95 2.2 2.4 3.69 1.79 3.71 5.96 2.83 1.1 0 2.77 3.5 1.1 4.23 0 1.79 1.61 4.09 2.83 3.04 3.04 2.83 2.4 3 1.39 0 0.69 1.39 0 2.71 3.85 2.56 0.69 3.89 3.18 2.94 1.95 2.89 3.69 4.66 0 3.81 1.79 1.61 2.71 0 2.08 3.14 0.69 2.89 0
21540 16810 7849 18598 8015 13761 11109 24322 10161 17678 13201 9730 25639 28324 17300 12638 39085 17509 5685 8818 11525 21347 8732 12848 17880 12070 3730 11645 10314 29767 16138 21948 11295 7652 13607 8795 9051 9744 12172 18590 10724 16722 10642 23761 8316 21975 12481 11180 19758 18950 16963 15338 14768 15140 20238 20221 27418 20045 20764 8066
M N M N M N M M M N M N N M N M M M M M M M M M M N N M M M N M N M M M M N N M N N N N M N N N M M N N N M N N N M N M
RT-total, mean reaction times for previous studies conducted at the Center for Research in Language; RT-target, reaction times for dominant responses only; Ln frequency, the log natural frequency for each word’s dominant response; VisComplexity, each picture’s objective visual complexity based on its picture Wle size in jpg format; Manip, the hand imagery classiWcation (M, manipulable, N, non-manipulable); (m) next to the picture name refers to items classiWed as involving mouth imagery, (f) refers to items involving foot; (¤) next to the item means item was removed from conservative analyses due to other body part involvement.
90
A. Arévalo et al. / Brain and Language 100 (2007) 79–94
Table 6 Action items No. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Picture name ¤
BARK (m) BEG¤ (m) BLOW¤ (m) BOX¤ (w) BRUSH CARRY¤ (w) CLIMB¤ (w) COUGH¤ (m) CRY¤ (m) CURL¤ (w) CURTSEY¤ (w) CUT DANCE DIP DRIP DRIVE DUST ERASE FALL FEED FLY FOLLOW¤ (w) FRIGHTEN¤ (w) GIVE GREET HUNT IRON KICK¤ (w, f) KNEEL LICK¤ (m) LIGHT LISTEN MAIL MOP POINT POP POUR PRAY¤ (w, m) PULL REACH¤ (w) READ RUN¤ (w, f) SALUTE SHOOT SHOWER SKI¤ (w) SLIDE SLIP¤ (w, f) SMILE¤ (m) SNOW SPREAD SQUEEZE SWEAT SWIM SWING¤ (w) TICKLE¤ (m) TIE VACUUM WAVE YELL¤ (m)
RT-total
RT-target
Ln frequency
VisComplexity
Manip
949 1348 1534 967 903 1253 1001 1334 962 1346 1306 1065 993 1317 980 999 1215 1319 1134 1241 914 1318 1322 1330 1216 1254 977 866 1331 1120 1298 1245 1246 1332 1102 1261 890 1224 1255 1300 993 912 1028 1032 974 1428 913 1238 1119 1266 1351 1133 1239 852 874 1258 1093 996 1224 1266
949 1292 974 963 888 1180 989 1255 934 1326 1203 1065 979 1294 947 989 1209 1244 1159 1208 914 1321 1246 1343 1174 1282 977 853 1252 1100 1304 1263 1134 1258 1063 1121 852 1216 1223 1261 993 918 1028 1012 947 1053 886 1231 1107 1221 1367 1128 1201 852 874 1172 1099 993 1207 1249
2.4 3.43 4.44 0.69 3.22 5.74 4.53 2.56 4.8 2.77 0.69 5.25 4.2 2.89 2.4 5.39 2.2 1.61 5.69 4.9 4.57 5.69 2.08 7.15 4.88 3.4 1.79 3.76 3.18 2.48 4.01 5.18 1.61 1.95 4.89 3 4.38 3.37 5.23 5.55 5.92 6.09 1.39 4.32 1.95 1.95 3.58 4.13 5.09 1.61 4.49 3.37 2.89 3.87 4.04 1.61 4.13 0.69 3.83 3.14
18031 17686 19790 16757 23911 17053 37429 33349 22897 27471 14133 18411 30516 20402 15971 35400 13403 23620 26229 22683 13178 19976 24409 27760 34427 45398 13323 17222 14002 18076 20907 37439 25541 20337 16800 15804 26916 45299 30784 18105 30065 17276 15575 19808 28383 17193 32449 27692 40153 44104 25846 17216 16947 16766 18530 18027 23682 30285 15853 20192
N M N N M N M M M M N M N M N M M M N M N N N M M M M N N M M N M M N N M N M N M M N M M M N N N N M M N N N M M M N N
RT-total, mean reaction times for previous studies conducted at the Center for Research in Language; RT-target, reaction times for dominant responses only; Ln frequency, the log natural frequency for each word’s dominant response; VisComplexity, each picture’s objective visual complexity based on its picture Wle size in jpg format; Manip, the hand imagery classiWcation (M, manipulable, N, non-manipulable); (m) next to the picture name refers to items classiWed as involving mouth imagery, (f) refers to items involving foot and (w) refers to items involving whole body imagery; (¤) next to the item means item was removed from conservative analyses due to other body part involvement.
A. Arévalo et al. / Brain and Language 100 (2007) 79–94
‘Manipulable noun’
« Look at this : (ACCORDION) »
‘Manipulable verb’
« I want to : (SQUEEZE) »
91
‘Non-manipulable noun’
« They saw the : (AIRPLANE) »
‘Non-manipulable verb’
« It started to : (SNOW) »
Fig. 7. Example of items from the PN task and the lead in sentences that preceded each: Manipulable Noun: ‘Look at this: ACCORDION.’; Non-Manipulable Noun: ‘They saw the: AIRPLANE.’; Manipulable Verb: ‘I want to SQUEEZE.’; Non-Manipulable Verb: ‘It started to : SNOW.’
both (see Table 1) also had a lesion in areas/pathways previously identiWed as ‘hand’ areas (following Tettamanti et al., 2005). We identiWed this area on our patients’ lesion reconstructions and through a visual comparison, we identiWed a lesion in this area for 10 out of 13 patients (77%). This can be more easily observed by using a lesion overlay method. For lesion overlays, lesion localization is initially veriWed by a board-certiWed neurologist who is blind as to the purpose of the study, using patients’ CT and/or MRI scans (obtained at least 5 weeks postonset). These lesions are reconstructed onto standardized axial templates, based on the atlas of DeArmond, Fusco, & Dewey (1976). These templates are then entered into a Macintosh PC, using inhouse lesion reconstruction software (Frey, Woods, Knight, Scabini, & Clayworth, 1987), which allows for lesion analysis. Using this technique we created a lesion overlay map for 13 patients who displayed the ‘manipulability’ eVect (a total of 16 out of 21 patients showed the eVect, yet reconstructed lesion maps were not available for three of these patients. The remaining scans are currently being acquired). Therefore, for these 13 patients, the lesion overlay map revealed that up to 60% of them had a lesion in a motor hand area/pathway region, visible on slice 9 (see Fig. 8). Although this could constitute some support for the idea of a correlation between lesion site and a deWcit in processing manipulable items in this type of behavioral task, we cannot conWrm this hypothesis without a larger patient sample. The second approach we used was to compare patients’ lesion volumes to their performance on this manipulability dimension. With information on lesion volumes for 15
Fig. 8. Lesion Overlay map for aphasic patients who displayed the ‘manipulability eVect’, i.e., their performance on manipulable items was at least 5 percentage points lower than their performance on non-manipulable items on PN, WRP or both. A total of 16 out of 21 patients showed the eVect. Reconstructed lesion maps were not available for three of these patients. For the patients included in this analysis, up to 60% of them had a lesion in a motor hand area/pathway (best seen on slice 9).
patients, we analyzed whether the manipulability eVect changed as a function of lesion volume (as opposed to lesion location or even aphasia type). Results revealed that larger lesions predicted a signiWcant ‘manipulability eVect’ (i.e., better processing of non-manipulable relative to manipulable items) both overall and in the WRP task independently. Again, these data are not conclusive, but they open the door to new inquiries in the study brain lesions, lexical and semantic processing which can be further probed in the future with larger patient samples. 4. Discussion Noun–verb (or action–object) diVerences have been widely studied and reported in the neuropsychological literature. The classic notion of a noun–verb double-dissociation in brain-injured patients has also been reported throughout the years, yet not without controversy. More recent studies have tested the notion that perhaps deeper, more semantically-based diVerences more accurately account for the discrepancy in performance between participants on these two lexical categories. Embodiment Theory (e.g., MacWhinney, 1999) suggests that the involvement of certain body parts in the processing of action or object concepts should activate the corresponding sensorimotor areas in the brain, whether an action is performed by the participant, performed by someone else and observed by the participant, or simply processed in the imagination. In this study we initially set out to test noun-verb diVerences and subsequently re-categorized our items to reXect diVerences in hand imagery (i.e., manipulable vs. non-manipulable items) by using data from our Gesture Norming study which codes which body parts are involved
92
A. Arévalo et al. / Brain and Language 100 (2007) 79–94
while producing gestures to the target words associated with these same items. Whereas the noun–verb diVerence did not provide support for a double-dissociation and resulted mainly in a single dissociation in only one of the tasks (PN), hand imagery, or ‘manipulability’, diVerences were more revealing: performance on these two categories dissociated doubly between patient and healthy groups in two of the three tasks, PN and WRP. SpeciWcally, whereas both control groups were signiWcantly more accurate at producing the manipulable items, patients displayed the opposite pattern: they were signiWcantly more accurate at processing the non-manipulable items. Another interesting Wnding was that while patients displayed stronger manipulability eVects on verbs relative to nouns, control groups showed a stronger eVect for nouns relative to verbs. When we then removed all items which could contain imagery for other body parts, such as the mouth (and/or tongue), foot (and/or leg) and whole body, the main eVects from these ‘conservative’ analyses did not diVer from the Wrst set of results. In addition, manipulability eVects became signiWcant for Broca’s patients in the reading (WR) condition, as did the interaction of Subgroup (Older controls vs. Patients) £ Manipulability for the WR condition (suggesting that controls and patients diverged in their performance on the third task as well; F(1, 1) D 4.7454, p < .0295). The fact that these eVects are seen in two of the three tasks (and in all three for the conservative analyses) suggests that these sensorimotor-based distinctions may be tapping into some deeper process which (unlike the noun– verb distinction) may hold up across tasks and sensory modalities. One possible interpretation is that patients’ lesions compromise at least part of their ability to process motor actions of the hand, either by directly aVecting areas located in motor cortex subserving hand motor movements or by compromising part of the network involved in such processing. In contrast, healthy participants may actually rely specially on such motor imagery when performing the same task. In other words, using such imagery to facilitate comprehension and naming may be a normal processing strategy, resulting in better performance on the ‘manipulable’ items. This discrepancy in strategies and/or abilities would therefore result in a double-dissociation between control and patient group performance on the present set of tasks. In addition, these strategies may be more signiWcant or useful in processing one word category vs. another (as seen by the diVerence in manipulability eVects for the two lexical categories in controls vs. patients) depending on the group, and in cases of brain injury, manipulability eVects may render one of these categories most vulnerable (e.g., verbs in the case of patients). Finally, this ‘manipulability eVect’ was seen for patients in all three aphasia groups. The only group which deviated from the ‘patient pattern’ in some analyses was the anomic group, the least severe of the three. However, this can probably not be attributed to aphasia severity, since this eVect was seen at the group level yet in fact, several of the patients showing some degree of ‘the
eVect’ were Anomics (6 out of 16, or 60% of all Anomics in this study). In addition, there were diVerent numbers of patients per group (10 Anomics, 6 Broca, 5 Wernicke), which limit the number of cross-group interpretations we could make. Our aim was clearly to test equal numbers of patients per group, yet this is often diYcult in studies involving severe patients (particularly in the case of studies requiring patients to verbally produce answers to relatively diYcult and varied stimuli). It is worth noting, however, that a theory that expects motor-type diVerences for diVerent aphasia types is a theory which assumes these same aphasia types (determined based on behavioral proWles measured with standardized tests) correspond to clean, anatomically distinct proWles. It is our view that aphasia types are useful for diagnostic practices, but we did not place any weight on them in terms of our expectations or hypotheses for this study. Instead, our aim was to test a hypothesis which places emphasis on the neurological capabilities of patients, which may or may not be reXected in their pre-assigned categories. In fact, patients showing the ‘manipulability eVect’ in this study belonged to all three aphasia type categories. The lesion overlay and the lesion volume analyses we conducted suggest two ways in which a possible connection of behavior to lesion can be further probed in the future, but no real conclusions can be reached at this early stage. The current behavioral Wndings, attained with a task originally designed to test one parameter (the classic noun–verb distinction), carry important implications for our understanding of the way we process semantic and lexical knowledge. They support theories of embodied cognition and provide insight into linguistic processing following brain injury as well as theories on aphasic classiWcations. This is also a Wrst glance into what we can gain from new tests of mental imagery expressed through gestures. Finally, two diVerent techniques using lesion information opened the door to possible future lines of analysis. Continuing work targeting these new parameters directly as well as using various imaging techniques is sure to further our knowledge in this exciting line of inquiry. Acknowledgments Support during the preparation of this work was provided by two separate grants: a Grant from the NIH Training Program in Cognitive Neurosciences (from the Institute in Neural Computation, University of California, San Diego) and Cross-linguistic Studies in Aphasia, (PHS 2 RO1 DC 0021618A1; previously under Bates and currently under Dronkers, PI/NIH/NIDCD 3 R01 DC00216-18A1). This work is dedicated to my late advisor, Dr. Elizabeth Bates. I thank Marco Tettamanti and Cristina Saccuman for helpful comments on earlier versions of the manuscript. A very special thank you goes to Juliana Baldo and Carl Ludy for all of their help on lesion reconstructions and patient information.
A. Arévalo et al. / Brain and Language 100 (2007) 79–94
References Allport, D. A. (1985). Distributed memory, modular systems and dysphasia. In S. K. Newman & R. Epstein (Eds.), Current perspectives in dysphasia. Churchill Livingstone, Edinburgh. Arévalo, A., Butler, A. C., Perani, D., Cappa, S., & Bates, E. (2005a). Introducing the Gesture Norming study: A tool for understanding on-line word and picture processing. Technical Report, CRL-0401. Center for Research in Language, University of California, San Diego. Arévalo, A., Moineau, S., Saygin, A., Ludy, C., & Bates, E. (2005b). In search of noun–verb dissociations in aphasia across three processing tasks. CRL Newsletter, 17(1) March. Bates, E., Andonova, E., D’Amico, S., Jacobsen, T., Kohnert, K., Lu, C-C., & Székely, A., et al., (2000). Introducing the CRL international picturenaming project (CRL-IPNP). Center for Research in Language Newsletter, 12(1). La Jolla: University of California San Diego. Bates, E., Burani, C., D’Amico, S., & Barca, L. (2001). Word reading and picture naming in Italian. Memory & Cognition, 29(7), 986–999. Bates, E., Chen, S., Tzeng, O., Li, P., & Opie, M. (1991). The noun–verb problem in Chinese aphasia (E. Bates, Ed.). Brain & Language, 41(2), 203–233. Bates, E., Wilson, S. M., Saygin, A. P., Dick, F., Sereno, M. I., Knight, R. T., et al. (2003). Voxel-based lesion-symptom mapping. Nature Neuroscience, 6(5), 448–450. Buccino, G., Binkofski, F., & Riggio, L. (2004). The mirror neuron system and action recognition. Brain and Language, 89, 370–376. Cappa, S. F., Binetti, G., Pezzini, A., Padovani, A., Rozzini, L., & Trabucchi, M. (1998b). Object and action naming in AD and frontotemporal dementia. Neurology, 50, 351–355. Caramazza, A., & Hillis, A. E. (1991). Lexical organization of nouns and verbs in the brain. Nature, 349, 788–790. Chen, S., & Bates, E. (1998). The dissociation between nouns and verbs in Broca’s and Wernicke’s aphasia: Wndings from Chinese. Special issue on Chinese aphasia. Aphasiology, 12(1), 5–36. Cohen, J. D., MacWhinney, B., Flatt, M., & Provost, J. (1993). PsyScope: an interactive graphic system for designing and controlling experiments in the psychology laboratory using Macintosh computers. Behavior Research Methods, Instruments & Computers, 25, 257–271. Cotelli, M., Borroni, B., Manenti, R., Alberici, A., Calabria, M., & Agosti, C., et al., (in press). Action and object processing in the fronto-temporal spectrum, Neuropsychology. Cree, G. S., & McRae, K. (2003). Analyzing the factors underlying the structure and computation of the meaning of chipmunk, cherry, chisel, cheese and cello (and many other such concrete nouns). Journal of Experimental Psychology: General, 132(2), 163–201. Daniele, A., Giustolisi, L., Silveri, M. C., Colosimo, C., & Gainotti, G. (1994). Evidence for a possible neuroanatomical basis for lexical processing of nouns and verbs. Neuropsychologia, 32(11), 1325–1341. DeArmond, S. J., Fusco, M., & Dewey, M. (1976). Structure of the human brain (2nd ed.). New York: Oxford University Press. Devlin, J. T., Moore, C. J., Mummery, C. J., Gorno-Tempini, M. L., Phillips, J. A., Noppeney, U., et al. (2002). Anatomic constraints on cognitive theories of category speciWcity. NeuroImage, 15, 675–685. Ehrsson, H. H., Geyer, S., & Naito, E. (2003). Imagery of voluntary movement of Wngers, toes, and tongue activates corresponding body-part-speciWc motor representations. Journal of Neurophysiology, 90, 3304–3316. Frey, R. T., Woods, D. L., Knight, R. T., Scabini, D., & Clayworth, C. (1987). DeWning functional areas with averaged CT sans. Society for Neuroscience Absracts, 13, 1266. Funnell, E. (2002). Semantic memory. In A. E. Hillis (Ed.), The Handbook of Adult Language Disorders. NY: Psych Press. Gerlach, C., Law, I., & Paulson, O. B. (2002). When action turns into words. Activation of motor-based knowledge during categorization of manipulable objects. Journal of Cognitive Neuroscience, 14(8), 1230–1239. Hauk, O., & Pulvermüller, F. (2004). Neurophysiological distinction of action words in the fronto-central cortex. Human Brain Mapping, 21, 191–201. Hauk, O., Johnsrude, I., & Pulvermüller, F. (2004). Somatotopic representation of action words in human motor and premotor cortex. Neuron, 41, 301–307.
93
Kellenbach, M. L., Brett, M., & Patterson, K. (2003). Actions speak louder than functions: the importance of manipulability and action in tool representation. Journal of Cognitive Neurosciences, 15(1), 30–45. Kertesz, A. (1979). Aphasia and associated disorders: Taxonomy, localization and recovery. New York: Grune & Stratton. Liu, H. (1996). Lexical access and diVerential processing in nouns and verbs in a second language. Unpublished doctoral dissertation, San Diego: University of California . MacWhinney, B. (1999). The emergence of grammar from embodiment. In B. MacWhinney (Ed.), The emergence of language (pp. 13–256). Mahwah, NJ: Lawrence Erlbaum. Martin, A., & Chao, L. L. (2001). Semantic memory and the brain: structure and processes. Current Opinion in Neurobiology, 11, 194–201. Martin, N., Dell, G. S., SaVran, E. M., & Schwartz, M. F. (1994). Origins of paraphasias in deep dysphasia: testing the consequences of a decay impairment to an interactive spreading activation model of lexical retrieval. Brain and Language, 47, 609–660. Pulvermüller, F., Hauk, O., Nikulin, V. V., & Ilmoniemi, R. J. (2005). Functional links between motor and language systems. European Journal of Neuroscience, 2, 793–797. Rizzolatti, G., Fadiga, L., Gallese, V., & Fogassi, L. (1996). Premotor cortex and the recognition of motor actions. Brain Research. Cognitive Brain Research, 3(2), 131–141. Rizzolatti, G., Fadiga, L., Matelli, M., Bettinardi, V., Paulesu, E., Perani, D., et al. (1996). Localization of grasp representations in humans by PET: 1. Observation versus execution. Experimental Brain Research, 111(2), 246–252. Saccuman, C., Perani, D., Bates, E.A., Danna, M., & Cappa, S.F. (2003). The neural correlates of noun and verb processing: semantic vs. grammatical eVects. Published abstract presented at the annual meeting for the Cognitive Neuroscience Society 2003, New York, NY. Székely, A., & Bates, E. (2000). Objective visual complexity as a variable in studies of picture naming. Center for Research in Language Newsletter. Székely, A., D’Amico, S., Devescovi, A., Federmeier, K., Herron, D., Iyer, G., et al. (2005). Timed action and object naming. Cortex, 41(1), 7–26. Tettamanti, M., Buccino, G., Saccuman, M. C., Gallese, V., Danna, M., Scifo, P., et al. (2005). Listening to action-related sentences activates frontoparietal motor circuits. Journal of Cognitive Neuroscience, 17(2), 273–281. Warrington, E. K., & McCarthy, R. (1987). Categories of knowledge: further fractionations and an attempted integration. Brain, 110, 1465–1473. Warrington, E. K., & Shallice, T. (1984). Category speciWc semantic impairments. Brain, 107, 829–854. Zingeser, L. B., & Sloan Berndt, R. (1990). Retrieval of nouns and verbs in agrammatism and anomia. Brain and Language, 39, 14–32.
Further reading Arévalo, A. (2002). Teasing apart actions and objects: a picture naming study. CRL Newsletter, 14(2). Bates, E., Bretherton, I., & Snyder, L. (1988). From Wrst words to grammar: Individual diVerences and dissociable mechanisms. New York: Cambridge University Press 326. Binkofski, F., Buccino, G., Posse, S., Seitz, R. J., Rizzolatti, G., & Freund, H. J. (1999). A fronto-parietal circuit for object manipulation in man: evidence from an fMRI study. European Journal of Neuroscience, 11, 3276–3286. Boroditsky, L., & Ramscar, M. (2002). The roles of body and mind in abstract thought. Psychological Science, 13(2) March. Cappa, S. F., Frugoni, M., Pasquali, P., Perani, D., & Zorat, F. (1998a). Category-speciWc naming impairment for artefacts: A new case. Neurocase, 4, 391–397. Cappa, S. F., & Perani, D. (2003). The neural correlates of noun and verb processing. Journal of Neurolinguistics, 16, 183–189. Cappa, S. F., Perani, D., Schnur, T., Tettamanti, M., & Fazio, F. (1998c). The eVects of semantic category and knowledge type on lexical-semantic access: a PET study. NeuroImage, 8, 350–359.
94
A. Arévalo et al. / Brain and Language 100 (2007) 79–94
Chatterjee, A. (2001). Language and space: some interactions. Trends in Cognitive Sciences, 5(2), 55–61. Clark, A. (2001). Reasons, robots and the extended mind. Mind & Language, 16(2), 121–145. Decety, J., Grèzes, J., Costes, N., Perani, D., Jeannerod, M., Procyk, E., et al. (1997). Brain activity during observation of actions. InXuence of action content and subject’s strategy. Brain, 120, 1763–1777. Federmeier, K., & Bates, E. (1997). Contexts that pack a punch: lexical class priming of picture naming. Center for Research in Language Newsletter, 12(2). La Jolla: University of California San Diego. Ferrari, P. F., Gallese, V., Rizzolatti, G., & Fogassi, L. (2003). Mirror neurons responding to the observation of ingestive and communicative mouth actions in the monkey ventral premotor cortex. European Journal of Neuroscience, 17, 1703–1714. Gentner, D. (1982). Why nouns are learned before verbs: linguistic relativity versus natural partitioning. In S. A. Kuczaj (Ed.), Language development: Language, thought and culture (Vol. 2, pp. 301–334). Hillsdale, NJ: Erlbaum. Gibbs, R. W., Jr. (2003). Embodied experience and linguistic meaning. Brain and Language, 84, 1–15. Glenberg, A. M., & Robertson, D. A. (2000). Symbol grounding and meaning: a comparison of high-dimensional and embodied theories of meaning. Journal of Memory & Language, 43, 379–401. Gopnik, A., & Choi, S. (1995). Names, relational words and cognitive development in English and Korean speakers: nouns are not always learned before verbs. In M. Tomasello & W. Merriman (Eds.), Beyond names for things: Young children’s acquisition of verbs. New Jersey: Erlbaum. Grafton, S. T., Arbib, M. A., Fadiga, L., & Rizzolatti, G. (1996). Localization of grasp representations in humans by positron emission tomography. 2. Observation compared with imagination. Experimental Brain Research, 112, 103–111. Grafton, S. T., Fadiga, L., Arbib, M. A., & Rizzolatti, G. (1997). Premotor cortex activation during observation and naming of familiar tools. NeuroImage, 6, 231–236. Hari, R., Forss, N., Avikainen, S., Kirveskari, E., Salenius, S., & Rizzolatti, G. (1998). Activation of human primary motor cortex during action observation: a neuromagnetic study. Proceedings of the National Academy of Sciences of the United States of America, 95, 15061–15065. Iyer, G. K. (2000). Picture naming in adults and children: an online behavioral study. Unpublished second year project report, San Diego: University of California. Johnson, S. H. (2000). Imagining the impossible: intact motor representations in hemiplegics. Neuroreport, 11(4), 729–732. Johnson, S. H., Corballis, P. M., & Gazzaniga, M. S. (2001). Within grasp but out of reach: evidence for a double-dissociation between imagined hand and arm movements in the left cerebral hemisphere. Neuropsychologia, 39, 36–50. Johnson, C. J., Paivio, A., & Clark, J. M. (1996). Cognitive components of picture naming. Psychological Bulletin, 120(1), 113–139. Jonkers, R., & Bastiaanse, R. (1998). How selective are selective word class deWcits? Two case studies of action and object naming. Aphasiology, 12, 193–206. Levelt, W. J. M. (1989). Speaking: From intention to articulation. Cambridge, MA: MIT Press. Levelt, W. J. M., Roelofs, A., & Meyer, A. S. (1999). A theory of lexical access in speech production. Behavioral & Brain Sciences, 22(1–38), 69–75. Luzzatti, C., Raggi, R., Zonca, G., Pistarini, C., Contardi, A., & Pinna, G. (2002). Verb–noun double dissociation in aphasic lexical impairments: the role of word frequency and imageability. Brain and Language, 81, 432–444.
Martin, A., Wiggs, C. L., Ungerleider, L. G., & Haxby, J. V. (1996). Neural correlates of category-speciWc knowledge. Nature, 379, 649–652. Menard, M. T., Kosslyn, S. M., Thompson, W. L., Alpert, N. M., & Rauch, S. L. (1996). Encoding words and pictures: a positron emission tomography study. Neuropsychologia, 34(3), 185–194. Molfese, D. L., Burger-Judisch, L. M., & Gill, L. A. (1996). Electrophysiological correlates of noun–verb processing in adults. Brain and Language, 54, 388–413. Murata, A., Fadiga, L., Fogassi, L., Gallese, V., Raos, V., & Rizzolatti, G. (1997). Object representation in the ventral premotor cortex (Area F5) of the monkey. Journal of Neurophysiology, 78, 2226–2230. Perani, D., Cappa, S. F., Bettinardi, V., Bressi, S., Gorno-Tempini, M., Matarrese, M., et al. (1995). DiVerent neural systems for the recognition of animals and man-made tools. Neuroreport, 6, 1637–1641. Perani, D., Cappa, S. F., Schnur, T., Tettamanti, M., Collina, S., Rosa, M. M., et al. (1999). The neural correlates of verb and noun processing: a PET study. Brain, 122, 2337–2344. Perani, D., Fazio, F., Borghese, N. A., Tettamanti, M., Ferrari, S., Decety, J., et al. (2001). DiVerent brain correlates for watching real and virtual hand actions. NeuroImage, 14, 749–758. Perani, D., Schnur, T., Tettamanti, C., Gorno-Tempini, M., Cappa, S. F., & Fazio, F. (1999). Word and picture matching: a PET study of semantic category eVects. Neuropsychologia, 37, 293–306. Price, C. J. (1998). The functional anatomy of word comprehension and production. Trends in Cognitive Sciences, 2(8), 281–287. Pulvermüller, F. (2005). Brain mechanisms linking language and action. Nature Review Neuroscience—July, 6(7), 576–582 Review. Pulvermüller, F., Preissl, H., Lutzenberger, W., & Birbaumer, N. (1996). Brain rhythms of language: nouns versus verbs. European Journal of Neuroscience, 8, 937–941. Ramnani, N., & Miall, R. C. (2004). A system in the human brain for predicting the actions of others. Nature Neuroscience, 7(1), 85–90 Rizzolatti, G., & Arbib, M. A. (1998). Language within our grasp. Trends in Neurosciences, 21, 188–194. Rizzolatti, G., Fogassi, L., & Gallese, V. (2001). Neurophysiological mechanisms underlying the understanding and imitation of action. Perspectives/Nature Reviews, 2, 661–670. Rizzolatti, G., Fogassi, L., & Gallese, V. (2002). Motor and cognitive functions of the ventral premotor cortex. Current Opinion in Neurobiology, 12, 149–154. Silveri, M. C., Gainotti, G., Perani, D., Cappelletti, J. Y., Carbone, G., & Fazio, F. (1996). Naming deWcit for non-living items: neuropsychological and PET study. Neuropsychologia, 35(3), 359–367. Sinha, C., & Jensen de López, K. (2000). Language, culture and the embodiment of spatial cognition. Cognitive Linguistics, 11(1/2), 17–41. Sirigu, A., Duhamel, J. R., Cohen, L., Pillon, B., Dubois, B., & Agid, Y. (1996). The mental representation of hand movements after parietal cortex damage. Science, New Series, 273(5281), 1564–1568. Snodgrass, J. C., & Vanderwart, M. (1980). A standardized set of 260 pictures: norms for name agreement, image agreement, familiarity, and visual complexity. Journal of Experimental Psychology: Human Learning and Memory, 6, 174–215. Tyler, L. K., Stamatakis, E. A., Dick, E., Bright, P., Fletcher, P., & Moss, H. (2003). Objects and their actions: evidence for a neurally distributed semantic system. NeuroImage, 18, 542–557. Vinson, D. P., & Vigliocco, G. (2002). A semantic analysis of grammatical class impairments: semantic representations of object nouns, action nouns and action verbs. Journal of Neurolinguistics, 15, 317–351. Vinson, D. P., Vigliocco, G., Cappa, S., & Siri, S. (2003). The breakdown of semantic knowledge: insights from a statistical model of meaning representation. Brain and Language, 86, 347–365.
