Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Neurocase 2003, Vol. 9, No. 5, pp. 421–435 1355-4794/03/0905–421$16.00 # Swets & Zeitlinger A Double Dissociation Between the Meanings of Action Verbs and Locative Prepositions David Kemmerer1,2 and Daniel Tranel2 1 Department of Audiology and Speech Sciences and Department of Psychological Sciences, Purdue University, West Lafayette, IN, USA, and 2Division of Cognitive Neuroscience, University of Iowa College of Medicine, Iowa City, IA, USA Abstract We describe two patients who manifested opposite patterns of performance on test batteries that evaluated production, comprehension, and semantic analysis of action verbs on the one hand (e.g. smile, wave, run) and locative prepositions on the other (e.g. in, on, over). JP failed all of the verb tests but passed all of the preposition tests, suggesting impaired knowledge of the meanings of action verbs but intact knowledge of the meanings of locative prepositions. In contrast, RR exhibited the reverse dissociation: he passed many of the verb tests but failed all of the preposition tests, suggesting mostly intact knowledge of the meanings of action verbs but impaired knowledge of the meanings of locative prepositions. This behavioral double dissociation reflects the fact that the two categories of words differ along several conceptual parameters. To a large extent, the patients exhibited a neuroanatomical double dissociation as well, since JP’s lesion is predominantly in the left frontal operculum whereas RR’s is predominantly in the left inferior parietal lobe and the posterior superior temporal region. This constitutes preliminary evidence that the meanings of action verbs and locative prepositions are represented by partially independent neural networks in the brain. Introduction Numerous studies have investigated the neural architecture of word meaning, and evidence from several approaches – the lesion method, functional imaging, and electrophysiology – has been converging on the view that lexicosemantic structures belonging to different conceptual domains are treated differently in the brain. Most of these studies have focused on nouns that refer to concrete entities such as animals, fruits/ vegetables, tools, persons, and body parts (see reviews by Martin and Chao, 2001; Forde and Humphreys, 2002). Comparatively less research has been devoted to verbs for physical actions (see reviews by Druks, 2002; Cappa and Perani, 2003), and even less attention has been paid to other word classes, such as prepositions that designate spatial locations (e.g. Kemmerer and Tranel, 2000a; Damasio et al., 2001). Here we present new experimental data suggesting that the meanings of action verbs and locative prepositions have partially independent neural bases. Specifically, we report two brain-damaged patients, one of whom has impaired knowledge of the semantic structures encoded by action verbs but intact knowledge of the semantic structures encoded by locative prepositions, and the other of whom has the opposite dissociation. Before describing the experiments, we review pertinent background material on the meanings and neural substrates of the two word classes. The meanings of action verbs and locative prepositions Although some verbs refer to abstract states and events (e.g. exist, remain, increase, elapse), the prototypical function of verbs in all languages is to denote physical actions, that is, situations in which an agent, such as a person or animal, engages in certain kinds of dynamic bodily movement. Most contemporary theories of verb semantics assume that causal structure constitutes the major skeletal framework around which verb meanings are built (e.g. Van Valin and LaPolla, 1997; Croft, 1998; Rappaport and Levin, 1998). The simplest verbs represent situations in which an agent performs a purely body-internal activity that does not necessarily bring about any changes in other entities (e.g. sing, laugh, wave, jog). Other verbs are more complex insofar as they express activities that do affect other entities in certain ways, such as by inducing a change of state (e.g. slice, engrave, wash, kill) or a change of location (e.g. pour, twist, load, smear). Correspondence to: David Kemmerer, Department of Audiology and Speech Sciences, 1353 Heavilon Hall, Purdue University, West Lafayette, IN 47907-1353, USA. Tel: þ1-765-494-3826; Fax: þ1-765-494-0771; e-mail: [email protected] 422 D. Kemmerer and D. Tranel In addition to the basic organizing factor of causal structure, the meanings of action verbs can be analyzed and compared in terms of the various semantic fields that they characterize. Levin (1993) sorted over 3,000 English verbs (the majority of which were action verbs) into approximately 50 classes and 200 subclasses. Representative classes include verbs of throwing (e.g. fling, hurl, lob, toss), verbs of contact by impact (e.g. pound, swat, tap, poke), verbs of creation (e.g. build, assemble, sculpt, weave), and verbs of ingesting (e.g. eat, chew, gobble, devour). The verbs in a given class collectively provide a richly detailed, multidimensional categorization of the relevant semantic field by making distinctions, sometimes of a remarkably fine-grained nature, along a number of different parameters. For instance, verbs of destruction are distinguished by the composition of the entity to be destroyed (e.g. tear vs. smash), the degree of force (e.g. tear vs. rip), and the extent of deformation (e.g. tear vs. shred). Similarly, for verbs of locomotion distinctions are made regarding manner of motion (e.g. skip, limp, tiptoe, march), rate of motion (e.g. walk, jog, run, sprint), and the attitude of the agent (e.g. sneak, strut, sashay, trudge). The verbs in a given class are also organized according to principled semantic relations such as the following (Fellbaum, 1998): synonymy, in which two verbs have nearly identical meanings (e.g. shout and yell); antonymy, in which two verbs have opposite meanings (e.g. lengthen and shorten); hyponymy, in which oneverb is at a higher taxonomic level than another (e.g. talk and lecture); and cohyponymy, in which two verbs are at roughly the same taxonomic level (e.g. bow and curtsey). A final point about the meanings of action verbs is that they vary considerably across languages. First, languages differ in the idiosyncractic semantic features of verbs (e.g. Newman, 1997, 2002). For instance, Wagiman (an Australian Aboriginal language) has an unusual verb (more precisely, a coverb), murr, which means ‘‘to walk along in the water looking for something with one’s feet’’ (Wilson, 1999). In addition, languages differ in the size of their verb lexicons. Some languages, like Yimas and Kalam (both in New Guinea), have very small inventories of only about 100 verb roots, so that verb serialization is required to express complex events. Thus, the conceptual event expressed in English by ‘‘I fetched some firewood’’ requires no less than six Kalam verb roots (Pawley, 1993): yad am mon p-wk d ap ay-p-yn 1SG go wood hit-break hold come put-PERF-1SG ‘‘I fetched some firewood’’ Another type of variation involves the encoding of motion events (e.g. Talmy, 1985; Wienold, 1995; Slobin, in press). In manner-incorporating languages (e.g. English, German, Russian, Swedish, Chinese), manner of motion is typically encoded by the verb (e.g. walk, saunter, creep, crawl), while path information is optionally expressed by prepositional phrases (e.g. across the street). By contrast, in path-incorporating languages (e.g. Modern Greek, Spanish, Japanese, Turkish, Hindi), the verb usually encodes direction of motion (e.g. Modern Greek: beno ‘‘move in’’, vgeno ‘‘move out’’, aneveno ‘‘move up’’, kateveno ‘‘move down’’), while manner information is optionally expressed by prepositional phrases or gerunds (e.g. me ta podia/perpatontas ‘‘on foot/walking’’). This brief review of crosslinguistic diversity raises the possibility that the meanings of action verbs are language-specific semantic structures, perhaps distinct from the kinds of nonlinguistic action concepts that are employed in behavioral planning, perceptual recognition, and inferential reasoning (Gennari et al., 2002, Papafragou et al., 2002). Action verbs and locative prepositions are similar insofar as they share the conceptual domain of space: physical actions necessarily unfold in space, and locations are also necessarily defined in terms of space. However, the two types of words differ semantically in several important ways. In contrast to action verbs, locative prepositions do not encode dynamic processes; instead, they designate static spatial relationships between objects. In the standard case, two objects are involved – the ‘‘figure’’, which is an object whose location is the focus of attention, and the ‘‘landmark’’, which is an object that serves as a point of reference for locating the figure. For example, in the sentence There’s a fly in your soup, the noun-phrase a fly specifies the figure, the noun-phrase your soup specifies the landmark, and the preposition in specifies the nature of the spatial relationship between them. Whereas action verbs number in the thousands and fall into a wide variety of classes and subclasses (at least in English), there are only a small number of locative prepositions (roughly 50 in English) and they comprise just two main classes – geometrical and projective (e.g. Herskovits, 1986; Garrod et al., 1999). Geometrical prepositions encode spatial relationships in terms of a combination of topological and functional features – for instance, in expresses ‘‘containment’’, on expresses both ‘‘contiguity’’ and ‘‘support,’’ around expresses ‘‘encirclement’’, and through expresses ‘‘penetration’’. Projective prepositions refer to spatial locations that are extended from the major dimensional axes of the landmark – thus, above and below designate relations of superiority and inferiority with respect to the vertical axis of the landmark, and in front of and in back of (or behind) designate relations of anteriority and posteriority with respect to the front/back axis of the landmark. Given that there are significantly fewer locative prepositions than action verbs, it is not surprising that the former words exhibit a much broader range of application than the latter (although, to be sure, verbs can also be quite polysemous). For example, on typically refers to a situation in which the figure is in direct contact with the landmark and is supported by it horizontally through gravity, such as a cup on a table, a child on a swing, a car on a road, etc. However, it is still possible to use on when one of the typical conditions is violated, as in the following situations: a cup on a table even though books and papers intervene (no direct contact between figure and landmark); a picture on a wall (vertical support by virtue of attachment instead of horizontal support by virtue of gravity); and a stripe on a shirt (the figure is not really supported by the landmark since it is part of the landmark). Double dissociation between verbs and prepositions Accounting for the remarkable flexibility of locative prepositions has proven to be a challenge for semantic analysis (Sandra and Rice, 1995). Yet another interesting difference between action verbs and locative prepositions involves the amount of detail that can be encoded in the semantic structures of the words. Whereas action verbs are capable of expressing very subtle nuances of meaning, locative prepositions are confined to expressing spatial relationships in terms of very schematic structural properties of the objects involved (e.g. Herskovits, 1986; Landau and Jackendoff, 1993; Lindstromberg, 1998). Usually the most important properties are general geometric features (e.g. volumes, surfaces, lines, points), axial structure (e.g. top/ bottom, front/back, left/right), and quantity (e.g. between requires two landmark objects, and among requires an aggregate). For the most part, metrical details are ignored, such as the particular sizes, shapes, and orientations of objects, or the precise distances between them (see Brown, 1994, for an important exception to this generalization). Finally, as with action verbs, the meanings of locative prepositions vary a great deal across languages. An instructive example is to look at how English, Finnish, Dutch, and Spanish describe the following situations: a cup on a table, an apple in a bowl, and a handle on a door (Bowerman, 1996; Bowerman and Choi, 2001). In English the same preposition, on, is used to refer to the ‘‘cup on table’’ and ‘‘handle on door’’ situations because both of them share the features of contact and support, and a different preposition, in, is used for the ‘‘apple in bowl’’ situation because it involves containment. Finnish is sensitive to a slightly different distinction. One case-marker, -ssa, specifies what might be called intimate contact, a semantic category that includes both containment and attachment, so it is used for both the ‘‘apple in bowl’’ and ‘‘handle on door’’ situations. A different case-marker, -lla, specifies non-intimate contact, which includes horizontal support through gravity, so it is used for the ‘‘cup on table’’ situation. As for Dutch and Spanish, they go in opposite semantic directions. Dutch opts for maximal segregation, employing three different prepositions – op, in, and aan – for the three situations, and Spanish opts for maximal generalization, using just one preposition – en – for the three situations. This crosslinguistic diversity suggests that when people talk about the spatial world, they must conceptualize their experience in ways that reflect the unique perspective captured by the inventory of expressions in the given language. Furthermore, this in turn suggests that the meanings of prepositions may be distinct from the kinds of spatial representations that are used for non-linguistic purposes such as perceptual categorization and motor control (Kemmerer, 1999; Kemmerer and Tranel, 2000a; Crawford et al., 2000; Munnich et al., 2001). The neural substrates of action verbs and locative prepositions Very little is currently known about the specific neural structures that underlie the meanings of action verbs and 423 locative prepositions. However, several recent studies (summarized below) suggest that, at the level of large-scale neural systems, the meanings of both types of words are subserved by various components of the so-called dorsal stream, predominantly in the left hemisphere. This is what one would expect, given that the dorsal stream has been alternately characterized as the ‘‘how’’ system and the ‘‘where’’ system, the term ‘‘how’’ deriving from evidence that the system plays an essential role in the visuomotor control of action (Goodale and Milner, 1992), and the term ‘‘where’’ deriving from evidence that the system also contributes to representing the locations of objects in space (Ungerleider and Mishkin, 1982). Damasio et al. (2001) report a PET study in which subjects named static pictures of actions with the appropriate verbs and also named static pictures of spatial relationships with the appropriate prepositions. Action naming recruited the left frontal operculum, the left inferior parietal region (supramarginal and angular gyri), and bilaterally the posterior-most aspect of the middle temporal gyrus and the bordering occipital cortex (i.e. the area known as MT – e.g. Dumoulin et al., 2000). The left frontal activation could reflect many aspects of verb processing, including the retrieval of phonological forms (Tranel et al., 2001; Hillis et al., 2002b), the retrieval of grammatical category information (Shapiro et al., 2001), and the retrieval of conceptual knowledge (McCarthy and Warrington, 1985; Daniele et al., 1994; Bak et al., 2001; Tranel et al., in press). Especially intriguing evidence for more widespread involvement of the left frontal lobe in representing the meanings of body-part-specific action verbs comes from recent studies suggesting that face-, arm-, and leg-related verbs (e.g. yawn, grab, and jump) are linked with the corresponding inferior, lateral, and superior sectors of somatotopically mapped premotor cortices (Buccino et al., 2001; Pulvermuller et al., 2001). Regarding the activation that Damasio et al. (2001) observed in the left inferior parietal region, it was significantly greater for naming actions performed with tools (e.g. write) than for naming actions performed without tools (e.g. walk), and other studies have also related this brain region to conceptual knowledge for tool manipulation and for various kinds of object-directed hand actions within peripersonal space (e.g. Binkofski et al., 1999; Chao and Martin, 2000; Buccino et al., 2001). As for the posterior middle temporal region (area MT), it is activated when subjects view static pictures of objects that imply motion, such as an athlete running versus an athlete standing still (Kourtzi and Kanwisher, 2000; Senior et al., 2000), and numerous studies have found this area (and the area just anterior and dorsal to it) to be associated with the meanings of action verbs (Wise et al., 1991; Martin et al., 1995; Fiez et al., 1996; Warburton et al., 1996; Kable et al., 2002; Tranel et al., in press). For the task of naming spatial relationships with prepositions, Damasio et al. (2001) found the largest and strongest activation to be in the left supramarginal gyrus; smaller and weaker activations were observed in the left frontal opercular 424 D. Kemmerer and D. Tranel and polar regions. It is not clear how each of these neural structures contributes functionally to the complex naming process; however, the results are consistent with lesion studies which have found that damage in the left frontoparietal territory is associated with impaired processing of locative prepositions, including specifically the meanings of these words (see also Friederici, 1982, 1985; Friederici et al., 1982; Grodzinsky, 1988; Tesak and Hummer, 1994; Kemmerer and Tranel, 2000a; Froud, 2001; Kemmerer, submitted; Tranel and Kemmerer, submitted). Furthermore, an fMRI study found activation in the parietal lobes bilaterally when subjects read sentences containing locative prepositions (Carpenter et al., 1999). It is also noteworthy that the left inferior parietal region has been linked with the representation of schematic categorical spatial relationships of the sort that are typically encoded by locative prepositions (e.g. Laeng, 1994; Kosslyn et al., 1998; Baciu et al., 1999). The current study As the preceding review indicates, action verbs and locative prepositions have similarities as well as differences at both semantic and neural levels of description. The two kinds of words are semantically similar because they both involve the conceptual domain of space, and this commonality is apparently reflected in closely related neural circuits in the left frontal opercular region and the left supramarginal gyrus. And yet the two kinds of words are also semantically distinct in many respects (e.g. dynamicity, causation, the range of semantic fields, and the degree of semantic specificity), and these differences in content may be correlated with differences in neural substrates, such as the relevance of the motion-related posterior middle temporal region for action verbs but not for locative prepositions. The neurocognitive differences between the meanings of action verbs and locative prepositions lead to the prediction that these two types of words could, at least in principle, be impaired independently of each other as a result of focal brain damage. This prediction was confirmed in two neurological patients who exhibited a double dissociation in their performance on two large batteries of tests, one focusing on multiple ways of processing the meanings of action verbs and the other focusing on multiple ways of processing the meanings of locative prepositions. Subjects Subject 1 JP is a fully right-handed (þ100 on the Geschwind-Oldfield scale) man who obtained a college education and worked as a drug store owner and operator until retiring around age 70. In 1987, at age 76, he suffered a left hemisphere infarction, which produced a severe aphasia and a mild right hemiparesis. On initial evaluation in the Benton Neuropsychology Laboratory conducted 10 days post-onset, JP demonstrated a severe non-fluent aphasia, most closely resembling a Broca-type, but he no longer had any hemiparesis, and his non-verbal abilities, including paralinguistic communication, were wellpreserved. We have studied JP on a regular basis ever since the onset of his condition. He has continued to manifest a severe expressive language impairment, which can be classified as a Broca-type aphasia, but his general cognitive abilities are otherwise well preserved. He is fully cooperative with all neuropsychological procedures, and he is readily capable of grasping task instructions provided they are given slowly with non-verbal accompaniments. In fact, he has excellent verbal comprehension in conversational speech formats, even though his performances on formal tests of aural comprehension tend to be impaired (see Table 1). Also, JP remains capable of singing very well. JP’s lesion is depicted in Fig. 1, which shows a threedimensional reconstruction with Brainvox (Frank et al., 1997) based on an MRI scan conducted 6/25/99. The lesion is in the heart of Broca’s area, in the left frontal operculum. The pars opercularis and pars triangularis are damaged, along with the underlying white matter. Cortically, the lesion extends anteriorly into the ventrolateral prefrontal region, in the pars orbitalis, and superiorly into the middle part of the premotor region (area 6). The anterior insula is damaged. However, the precentral gyrus and basal ganglia are minimally affected, which could explain why JP demonstrated virtually no right-sided motor defect. Posteriorly, there is no significant cortical involvement past the Rolandic sulcus; however, some of the white matter underneath the postcentral gyrus and inferior parietal lobule is affected. The temporal lobe appears to be entirely spared. Table 1 shows JP’s performance on standardized neuropsychological tests that are commonly used in the Benton Neuropsychology Laboratory and that give a comprehensive overview of his mental capacities (Tranel, 1996). These data were collected between 1999 and 2000, contemporaneously with the lesion data and with the experimental studies reported below. JP performed primarily within the average range on various subtests of the WAIS-III and the WRAT-III. His score on a test of non-verbal anterograde memory was normal. (The patient’s aphasia precluded valid administration of verbal IQ and memory tests.) Regarding his linguistic abilities, many subtests from the Multilingual Aphasia Examination and the Boston Diagnostic Aphasia Examination were administered, and he was impaired on all of them except for the two basic reading measures. His ability to comprehend written words was fairly wellpreserved, but his reading speed was defective. JP was also given several tests that evaluate visual object recognition and spoken naming across different conceptual categories (Tranel et al., 1997). The stimuli consisted of line drawings of objects, and accurate naming responses were accepted as correct identification. If naming was inaccurate or absent, the subject was prompted to generate specific, detailed descriptions of the stimuli. These were rated by experimenters to determine whether they conveyed sufficient information Double dissociation between verbs and prepositions 425 Table 1. Comprehensive neuropsychological evaluations of JP and RR JP Intellect and academic achievement Wechsler Adult Intelligence Scale-III (Scaled scores) Performance IQ Picture Completion Digit symbol-coding Block design Matrix reasoning Picture arrangement Wide range achievement test-III (Standard scores) Reading Non-verbal anterograde memory Benton visual retention test (correct/errors) Standardized assessment of language Multilingual aphasia examination (Percentiles) Controlled oral word association Visual naming Sentence repetition Token test Aural comprehension Reading comprehension Boston diagnostic aphasia examination Boston naming (#/60) Responsive naming (#/30) Reading (#/10) Iowa-Chapman speed of reading (#/25) Visual object recognition and spoken naming Animals (n ¼ 41) Recognition Naming (for the items correctly recognized) Fruits/Vegetables (n ¼ 32) Recognition Naming (for the items correctly recognized) Tools/Utensils (n ¼ 63) Recognition Naming (for the items correctly recognized) Vehicles (n ¼ 12) Recognition Naming (for the items correctly recognized) Musical instruments (n ¼ 12) Recognition Naming (for the items correctly recognized) RR Score Interpretation Score Interpretation 100 9 7 12 11 12 Average Average Low average High average average High average 132 16 10 12 17 17 Superior Superior Superior High average Superior Superior 93 Average 68 Borderline 6/6 Normal 8/3 Normal <1st %ile <1st %ile <1st %ile <1st %ile <1st %ile 59th %ile Defective Defective Defective Defective Defective Normal <1st %ile <1st %ile <1st %ile <1st %ile <1st %ile 59th %ile Defective Defective Defective Defective Defective Normal 12 24 7 4 Defective Defective Borderline Defective 9 6 9 10 Defective Defective Normal Defective 59% (z ¼ 11.8) 50% (z ¼ 14.7) Defective Defective 88% (z ¼ 1.4) 44% (z ¼ 16.7) Normal Defective 81% (z ¼ 3.0) 31% (17.1) Defective Defective 97% (z ¼ 1.1) 65% (z ¼ 7.9) Normal Defective 97% (z ¼ 0.2) 61% (9.3) Normal Defective 98% (z ¼ 0.5) 29% (z ¼ 18.0) Normal Defective 100% (z ¼ 0.8) 92% (z ¼ 2.7) Normal Defective 100% (z ¼ 0.8) 58% (z ¼ 19.7) Normal Defective Defective Defective 100% (z ¼ 1.1) 25% (z ¼ 16.0) Normal Defective 99th %ile 40th %ile 31 29 High normal Normal Normal Normal 83% (z ¼ 3.9) 40% (z ¼ 12.6) Visuospatial perception/construction Facial discrimination Judgment of line orientation Complex figure copy (#/36) Three-dimensional block construction (#/29) 32nd %ile 74th %ile 32 29 Normal High normal Normal Normal Executive function Wisconsin card sorting test Categories completed Perseverative errors 6 16 Normal Normal about the entity to support the scoring of the response as a correct identification, i.e. as adequate retrieval of conceptual knowledge. JP’s recognition was normal for the categories of tools/utensils and vehicles, but was impaired for the categories of animals, fruits/vegetables, and musical instruments. His naming was impaired for all of the categories. Turning to the assessment of visuospatial perception and construction, JP performed well on all of the tests that were administered. Finally, with regard to executive functions, it is 6 7 Normal Normal notable that he performed well on the Wisconsin Card Sorting Test, indicating preservation of higher-order reasoning and judgment capacities. Subject 2 RR is a fully right-handed (þ100) man who obtained a master’s degree in international affairs and worked for roughly 40 years in the printing and publishing business. 426 D. Kemmerer and D. Tranel Fig. 1. Lesion reconstruction for patient JP from a magnetic resonance imaging scan. The damage affects most of the cortex and white matter in the supramarginal and angular gyri, and the posterior part of the superior temporal gyrus; i.e. nearly all of what is usually demarcated as Wernicke’s area. There is an additional, smaller area of damage in the left prefrontal region, in the pars opercularis, as well as some damage in the white matter of the left temporal pole. Both the cortex and underlying white matter associated with the middle temporal gyrus, including the posterior extension of this region into the temporo-occipital transition zone, are undamaged. Initially he displayed global aphasia, but this gradually resolved. In fact, by the time he initially came to our laboratory in 1995, his general communication abilities were remarkably effective, even though his performance on formal language tests remained for the most part impaired, as described below. The results of RR’s general neuropsychological assessment are shown in Table 1. These data were obtained contemporaneously with both the lesion data and the experimental studies reported below. His performance IQ was well above average (132), and he obtained superior scores on many of the subtests of the WAIS-III. On the WRAT-III, however, he had borderline reading performance. His score on a test of non-verbal anterograde memory was normal. (As with JP, the patient’s aphasia precluded valid administration of verbal IQ and memory tests.) Regarding linguistic abilities, like JP he was impaired on all of the aphasia subtests except for the two basic reading measures. Also like JP, while his written comprehension was intact, his reading speed was poor. With respect to visual object recognition and spoken naming, RR exhibited a consistent pattern of normal recognition but impaired naming for all of the categories that were tested. It is noteworthy, however, that separate experiments revealed intact written naming for all of the categories (mean ¼ 92%, s.d. ¼ 6.5%; see Kemmerer et al., submitted). This constitutes further evidence that his conceptual knowledge of concrete entities is normal, and also indicates that his poor spoken naming is probably due to a deficit at one or more of the stages along the processing pathway that leads from meaning to articulation. Shifting to the domain of visuospatial perception/ construction, his scores were uniformly in the normal to highnormal range. Finally, his executive functions were wellpreserved, as indicated by normal performance on the Wisconsin Card Sorting Test. Experiments Involving action verbs Methods Fig. 2. Lesion reconstruction for patient RR from a magnetic resonance imaging scan. In January, 1991, at the age of 62, he suffered a left hemisphere infarction. The patient’s lesion is depicted in Fig. 2, which shows a three-dimensional reconstruction with Brainvox based on an MRI scan conducted 7/13/00. Six tests were used to evaluate the subjects’ ability to produce, comprehend, and semantically analyze action verbs. The design characteristics and processing requirements of the tests are described below. A complete list of the items used in the tests is provided by Kemmerer et al. (2001a), and further details appear in Fiez and Tranel (1997). [Some of the original tests described by Fiez and Tranel (1997) had a few more items than were used in the study by Kemmerer et al. Double dissociation between verbs and prepositions (2001a). Some items were omitted because of unreliable performances in normal control subjects. The slightly modified tests presented in Kemmerer et al. (2001a) are the same ones that were employed in the current study]. Naming test 100 color photographs of various actions are presented to the subject on a Caramate 4000 slide projector in free field. Each picture is intended to elicit a specific verb or else one of a small set of verbs that are all considered to be acceptable responses based on normative data. The first 75 items are single pictures that show ongoing activities (e.g. swimming), and the last 25 items are picture pairs that show both the initial and final states of completed events (e.g. peeled). The target verbs in this test (and also in the other tests) come from a wide range of semantic fields. The test requires recognizing the objects in the pictures as well as inferring what kind of action is taking place, i.e. activating the most appropriate action concept. Once an action concept has been activated, the verb semantic structure that corresponds most closely to it is selected, and then the verb’s phonological form is retrieved for production. Verb-picture matching test For each item, a written verb is presented to the subject together with color photographs of two different actions, and the subject’s task is to determine which action the verb describes. For example, in one item the verb is kicking and the two pictures show a person kicking a ball and a person rolling a ball. There are 69 items. The first 43 items contain single pictures of ongoing activities, and the last 26 contain picture pairs showing the initial and final states of completed events. Like the Naming test, this test requires complex visual processing of the pictorial stimuli; however, while the Naming test only has one picture (or picture pair) per item, the Matching test has two, so the visual processing demands are greater. On the other hand, although both the Naming test and the Matching test require activating the semantic structures of verbs, only the Naming test also requires voluntary retrieval of the phonological forms of verbs, so in this regard the Matching test may be somewhat easier than the Naming test. Picture attribute test and word attribute test The next two tests are similar insofar as they both emphasize a certain kind of analytic or inferential processing, specifically the ability to compare the values of two action concepts for a particular predetermined attribute. In the Picture Attribute test, the subject is shown two color photographs of actions (or two pairs of them) arranged on pages in a binder notebook, and is asked a question about which action satisfies a particular value for a single attribute. The test contains 72 items, 48 of which have single pictures of ongoing actions and 24 of which have picture pairs for completed events. The attribute questions are as follows: (1) Which action makes the loudest noise? (2) Which action would be most physically tiring? (3) 427 Which action would take more time to complete? (4) Which action would require a specified kind of movement (e.g. moving hands closer together, moving hands up/down, moving hands in a circle)? (5) Which action would be most enjoyable/harmful/helpful? (6) Which change of state was accomplished using a particular tool or utensil (e.g. knife, hammer)? (7) Which change of state was most permanent? (8) Which change of state best represented an improvement to the object? This test requires visual processing of the pictorial stimuli and activation of the appropriate action concepts. The main task of answering the attribute questions involves the following additional operations: decomposing the internal structure of each action concept, identifying the attribute at issue, determining its typical value, comparing the values for the different concepts, and making a decision about which one fulfills the target criterion. This test does not explicitly involve verbs in either the stimuli or the responses; nevertheless, it is likely that normal subjects make use of verbs when they perform the test, especially given the larger situational context of taking a series of tests that probe knowledge of verbs (see Gennari et al., 2002, for related experiments and discussion). The Word Attribute test is parallel in design to the Picture Attribute test, except the stimuli are written verbs instead of pictures. There are 62 items, 40 with verbs in progressive form and 22 with verbs in past tense form. With respect to processing requirements, the semantic structures of the verbs must be activated, and we assume that it is through these language-specific representations that more richly detailed action concepts are retrieved. Real-world knowledge about typical action scenarios is necessary in order to answer the attribute questions for many of the items. For instance, in one item the two verbs are singing and yawning, and the attribute question is which action makes the loudest sound. Although the answer is clearly singing, this reflects knowledge about how the two actions are typically executed, not how they are necessarily executed; after all, it is obviously possible for someone to sing very softly or yawn very loudly. Picture comparison test and word comparison test The last two tests are similar to the previous two tests in that they emphasize analytic/inferential processing, but they are unique in the specific nature of these processing requirements. In particular, both of these tests employ an ‘‘odd one out’’ paradigm in which the subject must compare three verb meanings and identify the one that is unrelated to the other two; moreover, the relevant criteria for comparison are not provided by the experimenter but must be discovered by the subject. The Picture Comparison test has 24 items, and each item consists of three photographs of ongoing actions arranged on pages in a binder notebook. The subject’s task, as alluded to above, is to indicate which picture shows a type of action that is somehow different from the other two. For example, in one item the three pictures show (1) a person wrapping a box with paper, (2) a person wrapping her wrist with a cloth, and (3) a person drying a plate with a towel. The visual processing 428 D. Kemmerer and D. Tranel demands for this test are substantial because three different pictures must be studied. In addition, a great deal of conceptual processing is required in order to determine which aspects of the depicted actions are most relevant for sorting them. As in the Picture Attribute Test, although verbs are not explicitly employed in either the stimuli or the responses, it is reasonable to assume that normal subjects sometimes use an implicit naming strategy to facilitate performance (again, see Gennari et al., 2002, for related experiments and discussion). Moreover, some of the items may depend on idiosyncratic aspects of the semantic structures of English verbs (Kemmerer et al., 2001a). The Word Comparison test is parallel in design to the Picture Comparison test, except the stimuli for each item consist of three verbs instead of three pictures. There are 44 items, and the subject’s task is to indicate which verb is somehow different in meaning from the other two. In each item, the two associated verbs have one of four types of semantic relation: synonymy, antonymy, hyponymy, and cohyponymy. In order to identify the appropriate relational dimension for grouping the verbs, the subject must retrieve and analyze the relevant semantic structures. As in the Word Attribute test, performance may be enhanced by also generating and inspecting mental images of particular action scenarios, since this could facilitate recognition of the features that are most important for the comparison process. Results and discussion JP’s and RR’s performances were evaluated relative to the following control data from Fiez and Tranel (1997): Naming (mean ¼ 85.0%, s.d. ¼ 5.0); Verb-Picture Matching (mean ¼ 92.1%, s.d. ¼ 4.6); Picture Attribute (mean ¼ 91.7%, s.d. ¼ 4.8); Word Attribute (mean ¼ 94.8%, s.d. ¼ 3.6); Picture Comparison (mean ¼ 83.6%, s.d. ¼ 8.3); Word Comparison (mean ¼ 88.7%, s.d. ¼ 8.1). [As noted earlier, some of the original tests described by Fiez and Tranel (1997) had a few more items than were used in the current study. The control data shown above reflect these adjustments.] For each test, a percent correct was calculated, and this was then converted to a z-score. Following standard convention, a subject was classified as impaired on a given test if the z-score was equal to or lower than -2.0 (see Damasio et al., in press). The results are presented in Table 2. Table 2. Results for tests involving action verbs JP Naming Matching Picture attribute Word attribute Picture comparison Word comparison RR % z % z 33 72 82 79 25 59 10.4 4.3 2.0 4.4 7.1 3.4 22 99 89 94 58 64 12.6 1.4 0.6 0.3 3.0 2.9 JP failed all six tests. His performance was extremely poor on both the Naming test and the Picture Comparison test, and his scores were also far below normal for the Verb-Picture Matching test, the Word Attribute test, and the Word Comparison test. His best performance was on the Picture Attribute test, but even here his score was still at the cutoff point for impairment. RR, on the other hand, exhibited a mixed pattern of performance across the six tests. He was impaired on the Naming test and also failed both of the Comparison tests, but he was nearly perfect on the Verb-Picture Matching test and was normal on both of the Attribute tests. Another way of portraying the differences between JP’s and RR’s performances is by directly comparing their scores. If we exclude the Naming test (on which both patients were severely impaired) and compute average scores for each patient on the remaining tests, it becomes clear that RR’s performance was much better than JP’s: RR’s mean percent correct was 80.8 whereas JP’s was 63.4, a difference of 17.4 points; and RR’s mean z-score was 1.08 (which is well within the normal range) whereas JP’s was 4.24 (which is well into the defective range), a difference of 3.16 points. Moreover, RR’s scores were consistently higher than JP’s on every test except Naming, which further accentuates RR’s superior performance relative to JP’s. In what follows, we look more closely at each patient’s performance profile, focusing first on JP and then shifting to RR. As noted in the earlier discussion of JP’s overall neuropsychological assessment, he has normal recognition of tools/ utensils and vehicles but impaired recognition of animals, fruits/vegetables, and musical instruments (Table 1). It is unlikely, however, that his category-related object recognition deficits contributed significantly to his failure on the tests of action verbs. Two of the tests – Word Attribute and Word Comparison – use only linguistic stimuli and hence do not require visual object recognition at all, yet JP still performed quite poorly. Moreover, even though the other four tests employ pictures as stimuli, only a very small number of items include objects from the categories that are challenging for JP; instead, the vast majority of items have people and tools/ utensils as the most important entities in the actions, and JP has normal recognition of these kinds of entities. These considerations suggest that JP’s low scores on the six tests stem from problems with verb processing rather than from category-related object recognition deficits. In the Methods section, we pointed out that the six tests are designed to measure verb knowledge from many different angles. In a previous study with a group of 89 neurological patients, we found that 30 of the patients were impaired on at least one of the tests and that they manifested a total of 22 distinct performance profiles, with each test dissociating from all of the others – a finding which constitutes experimental evidence that the tests do in fact require different processing operations that can be disrupted independently of each other (Kemmerer et al., 2001a, b). Regarding JP’s performance profile, it may be that he has a combination of impairments affecting several different aspects of verb processing. However, Double dissociation between verbs and prepositions given that he failed the entire set of tests, the simplest explanation is that, at the very least, he has a representational disorder affecting his knowledge of the semantic structures encoded by action verbs. Turning now to RR, as noted earlier he failed three of the tests (Naming, Picture Comparison, and Word Comparison) but passed the others (Verb-Picture Matching, Picture Attribute, and Word Attribute). This pattern suggests that his knowledge of verb meanings is still mostly intact, because otherwise he would not have been able to achieve such high scores on half of the tests in the battery. Additional evidence that he retains substantial knowledge of the semantic structures of action verbs comes from three other studies in which he performed within normal limits on forced-choice threealternative verb-picture matching tasks (Kemmerer, 2000a, 2003; Kemmerer and Wright, 2002). [He is referred to as 1962RR in those studies. JP did not participate in the studies, so we do not have data for him.] To fully account for RR’s performance profile, however, it is necessary to explain why he obtained low scores on certain tests. This is a complex issue, and below we elaborate and evaluate several different approaches to addressing it. First, in the three studies mentioned above (Kemmerer, 2000a, 2003; Kemmerer and Wright, 2002), RR consistently displayed robust dissociations between, on the one hand, preserved sensitivity to subtle, idiosyncratic semantic features of verbs that are irrelevant to grammar, and on the other hand, impaired sensitivity to aspects of linguistic meaning that strongly influence the interaction between verbs and various morphosyntactic constructions (see also Kemmerer, 2000b, for an analogous study involving prenominal adjective order). It is important to consider whether these findings can shed any light on RR’s performance profile in the current study. Space limitations preclude a summary of all of the studies, but we will attempt to convey a sense of the general pattern of results by reviewing the essential elements of just one study, namely the study that focused on the verbal un- prefixation construction (Kemmerer and Wright, 2002). This construction is directly associated with the schematic meaning ‘‘X causes Y to come out of a constricted spatial configuration relative to Z’’, and hence it only licenses verbs that express the creation of some kind of constricted spatial configuration which is potentially reversible (e.g. wrapunwrap, buckle-unbuckle, clog-unclog versus press- unpress, press, rotate- unrotate, submerge- unsubmerge). These semantic constraints are revealed in a rather striking way by the different uses of the verb cross: one can cross one’s arms and then uncross them (because a constricted spatial configuration is created and then reversed), but if one crosses a street and then walks back again, it would sound strange to say that one has uncrossed the street (because no constricted spatial configuration is involved). RR performed well on a verb-picture matching test that required him to discriminate between aspects of verb meaning that are completely irrelevant to un-prefixation (e.g. he correctly selected twisting instead of bending or folding to describe a picture of a woman 429 twisting a long thin sponge; note that all three verbs satisfy the general semantic criteria for un-prefixation but differ in the unique kinds of object manipulations that they encode). However, he failed a grammaticality judgment test that probed his knowledge of the particular semantic features that determine which verbs can occur in the construction (e.g. he incorrectly rated the verbs uncoil, unclip, and unshackle as being ungrammatical, and incorrectly rated the verbs unturn, undangle, and unfluff as being grammatical). A third test showed that his failure on the judgment test was not due to problems with various task demands, but was instead most likely due to a disturbance of his knowledge of the semantic constraints on un-prefixation. These findings (and analogous findings in the other studies cited above) suggest that RR has some kind of impairment of grammatically relevant aspects of linguistic meaning. The question at hand is whether this information can provide insight into RR’s performance profile across the six verb tests in the current study. While granting that the issues surrounding this question are quite complicated, we suspect that the answer is no, for the following two reasons. First, all of the tests in the verb battery depend far more on knowledge of grammatically irrelevant aspects of verb meaning than on grammatically relevant aspects, so there is no obvious reason why an impairment of the latter would cause RR to fail any of the tests, let alone just the Naming test and the two Comparison tests. Second, although it is possible that RR has an impairment of the grammatically relevant aspects of verb meanings, it is also possible that he does not. This is because his failure on the various grammaticality judgment tests that probe his sensitivity to the semantic factors underlying the interaction between verbs and certain morphosyntactic constructions could in principle reflect an impairment that largely spares verb meanings and instead affects constructional meanings (for elaboration of the distinction between verb meanings and constructional meanings, see Goldberg, 1995, 2003). [This uncertainty as to the exact nature of RR’s disorder of grammatical semantics is discussed in the papers cited above (Kemmerer, 2000a, p. 1014; Kemmerer and Wright, 2002, p. 421; Kemmerer, 2003, p. 29).] Another approach to explaining RR’s low scores on the Naming test and on the two Comparison tests is based on the fact that Fiez and Tranel’s (1997) control subjects performed worse on these three tests (with means between 83.6% and 88.7% and standard deviations between 5.0 and 8.3) than on the other three tests (with means between 91.7% and 94.8% and standard deviations between 3.6 and 4.8), suggesting differences in difficulty. Thus one might suppose that RR’s deficits only become apparent on the more challenging tests in the battery. An explanation along these lines could perhaps be applied to RR’s failure on the two Comparison tests because although his z-scores of 3.0 and 2.9 are plainly in the defective range, they are not dramatically low. However, such an explanation is less plausible for RR’s failure on the Naming test because his z-score of 12.6 is far out of proportion to the others. 430 D. Kemmerer and D. Tranel Yet another approach, which is an extension of the previous one and is our preferred interpretation of the data, is that RR has several distinct impairments that interfere with specific ways of processing the forms and meanings of verbs while leaving the semantic structures themselves more or less intact. Looking first at his failure on the two Comparison tests, this could derive from a moderately impaired ability to carry out one or more of the analytic operations that are shared by those tests but are not required by any of the other tests, such as identifying the particular semantic criteria that are relevant to sorting the three stimuli in each item (Kemmerer et al., 2001a). This approach to explaining RR’s performance is admittedly speculative, but it is much more precise than merely saying that the two Comparison tests are more ‘‘difficult’’ than the other tests. Regarding his failure on the Naming test, it is most likely due to a deficit involving lexical access and/or phonological planning. In our presentation of RR’s neuropsychological background, we pointed out that he has impaired spoken naming of various categories of concrete entities but preserved written naming of the very same entities. Given this context, it seems quite plausible that his low score on the Naming test in the verb battery reflects a disorder which affects the phonological production of not just nouns but also verbs. We did not test RR’s written naming of actions because at the time that the verb battery was administered, we had not yet systematically investigated his superior written over spoken naming of concrete entities; however, we predict that his written naming of actions would be within the normal range because, first, his knowledge of verb meanings appears to be mostly intact, and second, the available data suggest that his word production deficit is restricted to the phonological modality. This prediction is consistent with documented cases of patients who have better written than spoken production of verbs (e.g. Hillis and Caramazza, 1995; Hillis et al., 2002a). Unfortunately, however, the prediction cannot be tested because RR recently passed away. Overall, the essential point that we would like to emphasize is that even though RR is apparently deficient at processing the semantic and phonological structures of verbs in certain ways, his representational knowledge of verb meanings seems to be fairly well-preserved. This certainly appears to be the case for the grammatically irrelevant aspects of verb meanings, and it may also be true for the grammatically relevant aspects, although the actual status of this component of verb meanings remains ambiguous. red arrow, and the landmark is indicated by a green arrow. Based on normative data, each picture is associated with either a single preposition or with two very similar prepositions (e.g. in back of and behind). The number of pictures for each preposition is as follows: on (13), in (13), around (3), through (3), above/over (13), below/under (13), in front of (6), in back of/behind (6), beside/next to (6), and between (4). For many of these prepositions, prototypical as well as nonprototypical situations are illustrated in the pictures. Experiments involving locative prepositions Odd one out test The subject is shown 45 sets of three pictures, with each set organized in the same manner as in the Matching #2 test. For this test, though, the task is to determine which picture in each set shows a spatial relationship that is different from the others – a paradigm similar to the one used for the Picture Comparison Test and the Word Comparison Test in the battery for action verbs. The red and green arrows pointing to the figure and landmark objects are crucial because without them Methods Five tests were used to evaluate the subjects’ knowledge and processing of locative prepositions. The stimuli for the first four tests consist of 80 black-and-white photographs of real objects in various spatial relationships (for a complete description of the items used in these tests, see Kemmerer and Tranel, 2000a). In each picture the figure is indicated by a Naming test All 80 pictures are presented to the subject on a Caramate 4000 slide projector in free field. For each one, the subject is asked a question designed to elicit the preposition that best describes the depicted spatial relationship – e.g. ‘‘Where is the cap in relation to the chair?’’ (answer ¼ on). The processing operations required by the test include recognizing the objects shown in each picture, categorizing their spatial relationship according to the most appropriate prepositional semantic structure, and retrieving the phonological form of the preposition so it can be articulated. Matching #1 test The same 80 pictures are presented again in the same manner, only this time the subject is asked to choose which of three written prepositions best describes each situation – e.g. ‘‘The cap is in/on/beside the chair’’ (answer ¼ on). This test differs from the Naming Test insofar as it requires comprehension rather than production of prepositions. In addition, it has a linguistic emphasis since it requires comparing the visuospatial representation of each picture with the semantic structures of three different prepositions, and then selecting the best match. Matching #2 test The subject is shown 50 sets of three pictures, with each set arranged on a separate page in a binder notebook. For a given set, the task is to determine which picture best represents a particular written preposition. Thus, for one item the preposition is in and the three pictures show (1) a window above another window, (2) eggs in a carton, and (3) a boy on a swing. This test differs from the Matching #1 Test insofar as the emphasis here is more on visual than linguistic processing, since the subject must compare the representations of three different pictures with the semantic structure of a single preposition and then select the best match. Double dissociation between verbs and prepositions the subject would not be constrained in determining which objects, and hence which spatial relationships, to attend to. However, the subject cannot rely on just the placement of the arrows in performing the task but must focus on the objects that the arrows pointed to. Many of the items in this test require access to the crosslinguistically unique semantic structures of English prepositions. For example, one item has the following three pictures: (1) a design on the side of a coffee cup, (2) a boy on a tricycle, and (3) eggs in a carton. In order to determine that the first two pictures are similar whereas the third is the ‘‘odd one out’’, the subject must recognize that the first two pictures both represent spatial relationships that involve contact but not containment, whereas the third picture represents a spatial relationship that involves both of these features, especially the latter. These are the semantic features that distinguish on from in. If one did not take them into account but instead relied solely on nonlinguistic perceptual features to perform the task, it is unlikely that one would be able to arrive at the correct answer because there are so many perceptual dimensions along which the pictures could be compared. For instance, one could treat the first picture as the ‘‘odd one out’’ because it represents a spatial relationship in which the figure is literally part of the landmark, whereas the second and third pictures represent spatial relationships in which the figure and landmark are independently moveable objects. Verification test This test employs a different set of visual stimuli than the first four tests. The subject is shown 44 line drawings of abstract shapes in various spatial relationships. In each picture, one shape is dark and the other light. The subject’s task is to determine whether the meaning of a particular written preposition correctly describes the location of the dark object relative to the light one. The number of pictures representing each preposition is as follows: on (6), in (6), around (4), across (4), above/over (6), below/under (6), next to/beside (6), and between (6). Since the shapes do not represent real objects, the visual processing of object-internal features is minimized while that of inter-object spatial relationships is maximized. Results and discussion JP’s and RR’s performances were evaluated relative to a control group of 10 neurologically and psychiatrically healthy subjects with the following demographic characteristics: male/female gender ratio (5/5); age (mean ¼ 50.2, s.d. ¼ 8.6); education (mean ¼ 13.6, s.d. ¼ 2.2); handedness (all right-handed). The control subjects performed well on all of the tests: Naming (mean ¼ 91.0%, s.d. ¼ 4.6); Matching #1 (mean ¼ 92.4%, s.d. ¼ 4.1); Matching #2 (mean ¼ 97.6%, s.d. ¼ 2.3); Odd One Out (mean ¼ 94.7%, s.d. ¼ 3.6); Verification (mean ¼ 91.0%, s.d. ¼ 4.3). JP’s and RR’s data were analyzed by first calculating the percent correct for each test, and then converting the percentage scores to z-scores. As with 431 Table 3. Results for tests involving locative prepositions JP Naming Matching #1 Matching #2 Odd one out Verification RR % z % z 93 94 100 100 – .4 .4 1.0 1.5 – 68 70 74 62 72 5.0 5.5 10.3 9.1 4.4 the tests for action verbs, the cutoff for impairment on each test was set at a z-score of 2.0. The results are shown in Table 3. JP performed extremely well on the Naming test, the two Matching tests, and the Odd One Out test. We were unable to administer the Verification test to him because it was not available during the experimental session, but it is likely that he would have passed it without difficulty since the extant data indicate unambiguously that his knowledge of the meanings of locative prepositions is fully intact. In contrast, RR failed all of the tests, consistently obtaining scores that were many standard deviations below the mean for the control subjects. This outcome strongly suggests that his knowledge of the meanings of locative prepositions is disrupted. This view is further supported by an analysis of the distribution of his errors, as shown in Table 4. The results clearly indicate that he was impaired on geometrical as well as projective prepositions in all of the different testing formats. Thus the disruption of his knowledge of the semantic structures encoded by locative prepositions appears to be quite broad. It is possible that RR’s failure on the Naming test was exacerbated by difficulty retrieving the phonological forms of prepositions. This hypothesis derives from data and arguments presented earlier which suggest that he has a spoken word production deficit for both nouns and verbs. It is also important to note, however, that RR’s failure on the other tests is probably not due to an impairment of just the forms, as opposed to the meanings, of prepositions. In both of the Matching tests as well as in the Verification test, prepositions are presented in written format (and are also read aloud by the experimenter), and RR’s orthographic input lexicon appears to be intact, based on his good performance on standardized reading tests (Table 1). Furthermore, although the Odd One Out test depends on knowledge of the language-specific semantic structures of prepositions, it does not require overt processing of the forms of prepositions in either the stimuli or the responses. Another important point about RR’s performance profile is that there is no evidence that his failure on the entire battery of preposition tests might reflect some kind of non-linguistic deficit involving the perceptual processing of the visuospatial stimuli. Independent testing demonstrated that his object recognition abilities are normal (Table 1), and anyhow the 432 D. Kemmerer and D. Tranel Table 4. Distribution of RR’s errors in the tests involving locative prepositions. Cells indicate the number (and percentage) of items that were answered correctly out of the total possible Preposition Naming on in around through across above/over below/under in front of in back of beside between 9/13 9/13 1/3 1/3 8/13 10/13 4/6 5/6 5/6 2/4 8/13 10/13 2/3 2/3 – 7/13 12/13 4/6 6/6 4/6 2/4 Total 54/80 (68%) 56/80 (70%) – Matching #1 Matching #2 Odd one out 3/8 6/8 3/3 0/3 8/10 7/10 1/2 0/2 – Verification 5/6 3/6 3/4 – – 8/8 7/8 3/4 4/4 3/4 3/6 4/6 2/3 1/3 2/3 – – 37/50 (74%) 28/45 (62%) Verification test does not even require object recognition. Furthermore, independent testing demonstrated that his visuospatial processing abilities are also normal (Table 1). This is especially significant in the present context because two of the visuospatial tests – specifically, Complex Figure Copy and Three-Dimensional Block Construction – assess nonlinguistic processing of the same general kinds of categorical spatial relationships that are encoded by prepositions. As we mentioned in the Introduction, however, the meanings of prepositions may be representationally autonomous, one reason being that there are substantial crosslinguistic differences in the particular categorical spatial distinctions made by prepositions, which suggests that when people talk about the spatial world, they must package their experience in ways that reflect the unique perspective captured by the inventory of prepositions in the given language. The notion that the semantic structures of prepositions are independent, language-specific mental representations is supported by RR’s dissociation between, on the one hand, impaired processing of categorical spatial relationships that are encoded by prepositions, and on the other hand, intact processing of categorical spatial relationships that are not encoded by prepositions but are instead used for non-linguistic perceptual and constructional tasks. The same type of dissociation, as well as the opposite type of dissociation, was found in other brain-damaged patients described by Kemmerer and Tranel (2000a). The implications of this double dissociation are discussed in depth in that paper (see also Kemmerer, submitted, for another case similar to RR). General discussion The two patients described in this study manifested opposite patterns of performance on test batteries that evaluated production, comprehension, and semantic analysis of action verbs on the one hand and locative prepositions on the other. Because the combined results constitute a double dissociation, it is impossible to account for the patients’ differential performance in terms of differential difficulty of the test batteries. Instead, the 2/4 4/6 5/6 – – 6/6 4/6 Total 33/50 35/50 10/15 3/11 2/4 30/46 38/46 13/19 16/19 20/25 8/14 (66%) (70%) (75%) (27%) (50%) (65%) (83%) (68%) (84%) (80%) (57%) 32/44 (72%) double dissociation is a powerful indicator that the meanings of action verbs and locative prepositions involve distinct mental representations and/or computations that can be independently disrupted by brain damage. JP failed the entire set of tests for verbs but performed extremely well on the tests for prepositions. This robust dissociation can be explained in terms of a representational disorder that affects his knowledge of the meanings of action verbs but spares his knowledge of the meanings of locative prepositions. In contrast, RR exhibited the reverse dissociation, with worse performance on prepositions than on verbs. His scores for the preposition tests were consistently in the defective range, and his errors extended to all of the prepositions that were used, suggesting that his knowledge of the meanings of these words is severely disrupted. However, he performed remarkably well on three of the tests in the verb battery, and he could only have gotten such high scores if his knowledge of the meanings of action verbs were mostly intact. He failed the remaining tests in the verb battery, but we argued that the most plausible explanation of his low scores on those tests is that they reflect impairments of specific processing operations, such as discovering the appropriate semantic criteria for sorting the stimuli (which is required by the two Comparison tests) and accessing the phonological forms of verbs (which is required by the Naming Test). As indicated in the Introduction, action verbs and locative prepositions differ along several conceptual parameters. Action verbs encode dynamic events; many of them refer to complex causal sequences; they are organized into hundreds of classes and subclasses, each devoted to characterizing a particular semantic field; and they often specify remarkably precise, fine-grained nuances of meaning. Locative prepositions, on the other hand, designate static situations; they never imply causation; they constitute a small inventory of words with very little internal organization into classes and subclasses; and the meanings that they express are distinctive for being quite schematic and coarse-grained. The discovery that these two categories of words dissociate from each other in brain-damaged patients is therefore in accord Double dissociation between verbs and prepositions with the increasingly well-supported notion that lexicosemantic structures with different conceptual properties tend to have different neural substrates (see reviews by Forde and Humphreys, 2002; Martin and Chao, 2001). JP and RR exhibited not only a linguistic double dissociation, but also to a large extent a corresponding neuroanatomical double dissociation. JP’s lesion is mostly restricted to the left inferior and middle premotor/prefrontal region, and this is one of the major areas to have been associated with the meanings of action verbs (McCarthy and Warrington, 1985; Daniele et al., 1994; Bak et al., 2001; Pulvermuller et al., 2001; Tranel et al., in press). The correlation between JP’s verb deficit and his premotor/prefrontal lesion provides further evidence for the ‘‘direct matching hypothesis’’, which holds that ‘‘we understand actions when we map the visual representation of the observed action onto our motor representation of the same action’’ (Rizzolatti et al., 2001, p. 664; see also Damasio, 1989; Stamenov and Gallese, 2002). RR’s lesion, on the other hand, is primarily in the left inferior parietal lobe and the left posterior superior temporal region; most importantly, his lesion is centered in the supramarginal gyrus, which has been associated with the meanings of locative prepositions (Kemmerer and Tranel, 2000a; Damasio et al., 2001; Kemmerer, submitted; Tranel and Kemmerer, submitted). As noted earlier, the middle temporal gyrus, including its extension posteriorly into the temporooccipital transition zone, is not damaged. These aspects of the relevant brain-behavior relationships are fairly clear, but other aspects are more subtle. Although RR’s lesion is concentrated in specific sectors of the left parietal and temporal lobes, he also has a small amount of damage in the left frontal operculum. It is apparently not large enough to severely disrupt his knowledge of the meanings of action verbs, but it may partially explain why he failed the Naming test as well as the two Comparison tests in the verb battery. Regarding the Naming test, the left frontal operculum (especially the underlying white matter) may contribute to retrieving the phonological forms of verbs (Tranel et al., 2001; Hillis et al., 2002b), so it is possible that RR’s lesion impaired this retrieval process. As for the two Comparison tests, several recent studies have implicated the left inferior prefrontal cortex (especially Brodmann’s areas 44 and 47) in various kinds of semantic working memory tasks that involve accessing, maintaining, monitoring and manipulating semantic representations stored elsewhere (e.g. Poldrack et al., 1999; Thompson-Schill et al., 1999; Devlin et al., 2003). Perhaps this region subserves certain semantic computations that are necessary for the Comparison tests but not for any of the tests that RR passed. A problem with this account, however, is that many of the patients in the group study reported by Kemmerer et al. (2001a) had damage in the left inferior prefrontal cortex but nevertheless performed within normal limits on one or both of the Comparison tests. And more generally, the fact that the 30 patients described in that study exhibited 22 distinct performance profiles across the six tests in the verb battery suggests that each test has certain 433 unique processing requirements that can be independently impaired, which in turn implies that the neural basis of semantic working memory is probably quite complex. More research is clearly needed to explore this topic further. Finally, it is interesting that neither patient has damage to the region known as MT, which generally is thought to include the transition zone between the posterior-most aspect of the middle temporal gyrus and the bordering occipital cortex (e.g. Dumoulin et al., 2000). This region has been associated with action verbs, specifically the component of meaning that involves visual motion patterns (Wise et al., 1991; Martin et al., 1995; Fiez et al., 1996; Warburton et al., 1996; Kable et al., 2002; Tranel et al., in press). If the findings of the present study are integrated with the accumulating research on area MT, the picture that emerges is one in which this area (and the area just anterior and dorsal to it; see Kable et al., 2002) plays an important but not sufficient role in the anatomically distributed neural system underlying verb semantics. In particular, RR’s fairly well-preserved knowledge of action verbs may be supported by a close interaction between area MT and the left premotor/prefrontal cortex (most of which is still intact for him, as discussed above); yet for JP, a fully functioning area MT is apparently not enough to sustain knowledge of action verbs when there is extensive damage to the left premotor/prefrontal cortex (which is consistent with the ‘‘direct matching hypothesis’’ mentioned earlier). The intricacies of this neural circuitry will undoubtedly become clearer as more research is done. Acknowledgement This work was supported by NINDS Program Project Grant NS19632. References Baciu M, Koenig O, Vernier MP, Bedoin N, Rubin C, Segebarth C. Categorical and coordinate spatial relations: fMRI evidence for hemispheric specialization. NeuroReport 1999; 10: 1373–8. Bak TH, O’Donovan DG, Xuereb JH, Boniface S, Hodges JR. Selective impairment of verb processing associated with pathological changes in Brodmann areas 44 and 45 in the motor neurone disease-dementia-aphasia syndrome. Brain 2001; 124: 13–20. Binkofski F, Buccino G, Posse S, Seitz RJ, Rizzolatti G, Freund HJ. A frontoparietal circuit for object manipulation in man: Evidence from an fMRIstudy. European Journal of Neuroscience 1999; 11: 3276–86. Bowerman M. The origins of children’s spatial semantic categories: Cognitive versus linguistic determinants. In: Gumpertz J, Levinson S, editors. Rethinking linguistic relativity. Cambridge: Cambridge University Press, 1996: 145–76. Bowerman M, Choi S. Shaping meanings for language: Universal and language-specific in the acquisition of spatial semantic categories. In: Bowerman M, Levinson S, editors. Language acquisition and conceptual development. Cambridge: Cambridge University Press, 2001: 475–511. Brown P. The INs and ONs of Tzeltal locative expressions: The semantics of static descriptions of location. Linguistics 1994; 32: 743–90. Buccino G, Binkofski F, Fink GR, Fadiga L, Fogassi L, Gallese V, Seitz RJ, Zilles K, Rizzolatti G, Freund HJ. Action observation activates premotor and parietal areas in a somatotopic manner: An fMRI study. European Journal of Neuroscience 2001; 13: 400–4. Cappa SF, Perani D. The neural correlates of noun and verb processing. Journal of Neurolinguistics 2003; 2/3: 183–9. 434 D. Kemmerer and D. Tranel Carpenter PA, Just MA, Keller TA, Eddy WF, Thulborn KR. Time-course of fMRI activation in language and spatial networks during sentence comprehension. NeuroImage 1999; 10: 216–24. Chao LL, Martin A. Representation of manipulable man-made objects in the dorsal stream. NeuroImage 2000; 12: 478–84. Crawford LE, Regier T, Huttenlocher J. Linguistic and non-linguistic spatial categorization. Cognition 2000; 75: 209–35. Croft W. Event structure in argument linking. In: Butt M, Geuder W, editors. The projection of arguments: Lexical and compositional factors. Stanford: CSLI, 1998: 21–64. Damasio AR. Time-locked multiregional retroactivation: A systems-level proposal for the neural substrates of recall and recognition. Cognition 1989; 33: 25–62. Damasio H, Grabowski TJ, Tranel D, Ponto LLB, Hichwa RD, Damasio AR. Neural correlates of naming actions and of naming spatial relations. NeuroImage 2001; 13: 1053–64. Damasio H, Tranel D, Grabowski TJ, Adolphs R, Damasio AR. Neural systems behind word and concept retrieval. Cognition; in press. Daniele A, Giustolisi L, Silveri M, Colosimo C, Gainotti G. Evidence for a possible neuroanatomical basis for lexical processing of nouns and verbs. Neuropsychologia 1994; 32: 1325–41. Devlin JT, Matthews PM, Rushworth MFS. Semantic processing in the left inferior prefrontal cortex: A combined functional magnetic resonance imaging and transcranial magnetic stimulation study. Journal of Cognitive Neuroscience 2003; 15: 71–84. Druks J. Verbs and nouns: A review of the literature. Journal of Neurolinguistics 2002; 15: 289–315. Dumoulin SO, Bittar RG, Kabani NJ, Baker CL, LeGoualher G, Pike GB, Evans AC. A new anatomical landmark for reliable identification of human area V5/MT: A quantitative analysis of sulcal patterning. Cerebral Cortex 2000; 10: 454–63. Fellbaum C. A semantic network of English verbs. In: Fellbaum C, editor. Wordnet. Cambridge: MIT Press, 1998: 69–104. Fiez JA, Raichle ME, Balota DA, Tallal P, Petersen S. PET activation of posterior temporal regions during auditory word presentation and verb generation. Cerebral Cortex 1996; 6: 1–10. Fiez JA, Tranel D. Standardized stimuli and procedures for investigating the retrieval of lexical and conceptual knowledge for actions. Memory and Cognition 1997; 25: 543–69. Forde EME, Humphreys GW, editors. Category specificity in brain and mind. New York: Psychology Press, 2002. Frank RJ, Damasio H, Grabowski TJ. Brainvox: An interactive, multimodal, visualization and analysis system for neuroanatomical imaging. NeuroImage 1997; 5: 13–30. Friederici A. Syntactic and semantic processes in aphasic deficits: The availability of prepositions. Brain and Language 1982; 15: 249–58. Friederici A. Levels of processing and vocabulary types: Evidence from on-line processing in normals and agrammatics. Cognition 1985; 19: 133–66. Friederici A, Garrett M, Schönle P. Syntactically and semantically based computations. Cortex 1982; 18: 525–34. Froud K. Prepositions and the lexical/functional divide: Aphasic evidence. Lingua 2001; 111: 1–28. Garrod S, Ferrier G, Campbell S. In and on: Investigating the functional geometry of spatial prepositions. Cognition 1999; 72: 167–89. Gennari SP, Sloman SA, Malt BC, Fitch WT. Motion events in language and cognition. Cognition 2002; 83: 49–79. Goldberg A. Constructions: A construction grammar approach to argument structure. Chicago: Univerisity of Chicago Press, 1995. Goldberg A. Constructions: A new theoretical approach to language. Trends in Cognitive Sciences 2003; 7: 219–24. Goodale MA, Milner AD. Separate visual pathways for perception and action. Trends in Neurosciences 1992; 15: 20–5. Grodzinsky Y. Syntactic representations in agrammatic aphasia: The case of prepositions. Language and Speech 1988; 31: 115–34. Herskovits A. Language and spatial cognition. Cambridge: Cambridge University Press, 1986. Hillis AE, Caramazza A. The representation of grammatical categories of words in the brain. Journal of Cognitive Neuroscience 1995; 7: 396–407. Hillis AE, Tuffiash E, Caramazza A. Modality-specific deterioration in naming verbs in nonfluent primary progressive aphasia. Journal of Cognitive Neuroscience 2002a; 14: 1099–1108. Hillis AE, Tuffiash E, Wityk RJ, Barker PB. Regions of neural dysfunction associated with impaired naming of actions and objects in acute stroke. Cognitive Neuropsychology 2002b; 19: 523–34. Kable JW, Lease-Spellmeyer J, Chatterjee A. Neural substrates of action event knowledge. Journal of Cognitive Neuroscience 2002; 14: 795–805. Kemmerer D. ‘‘Near’’ and ‘‘far’’ in language and perception. Cognition 1999; 73: 35–63. Kemmerer D. Grammatically relevant and grammatically irrelevant features of verb meaning can be independently impaired. Aphasiology 2000a; 14: 997–1020. Kemmerer D. Selective impairment of knowledge underlying prenominal adjective order: Evidence for the autonomy of grammatical semantics. Journal of Neurolinguistics 2000b; 13: 57–82. Kemmerer D. Why can you hit someone on the arm but not break someone on the arm? A neuropsychological investigation of the English body-part possessor ascension construction. Journal of Neurolinguistics 2003; 16: 13–36. Kemmerer D. A neuropsychological dissociation between the spatial and temporal meanings of English prepositions. Manuscript under review. Kemmerer D, Tranel D, Manzel K. An exaggerated effect for proper nouns in a case of superior written over spoken word production: Implications for the organization of the modality-specific output lexicons. Manuscript under review. Kemmerer D, Tranel D. A double dissociation between linguistic and perceptual representations of spatial relationships. Cognitive Neuropsychology 2000a; 17: 393–414. Kemmerer D, Tranel D. Verb retrieval in brain-damaged subjects: II. Analysis of errors. Brain and Language 2000b; 73: 393–420. Kemmerer D, Tranel D, Barrash J. Patterns of dissociation in the processing of verb meanings in brain-damaged subjects. Language and Cognitive Processes 2001a; 16: 1–34. Kemmerer D, Tranel D, Barrash J. Addendum to ‘‘Patterns of dissociation in the processing of verb meanings in brain-damaged subjects’’. Language and Cognitive Processes 2001b; 16: 461–3. Kemmerer D, Wright SK. Selective impairment of knowledge underlying unprefixation: Further evidence for the autonomy of grammatical semantics. Journal of Neurolinguistics 2002; 15: 403–32. Kosslyn SM, Thompson WL, Gitelman DR, Alpert NM. Neural systems that encode categorical versus coordinate spatial relations: PET investigations. Psychobiology 1998; 26: 333–47. Kourtzi Z, Kanwisher N. Activation in human MT/MST by static images with implied motion. Journal of Cognitive Neuroscience 2000; 12: 48–55. Laeng B. Lateralization of categorical and coordinate spatial functions: A study of unilateral stroke patients. Journal of Cognitive Neuroscience 1994; 6: 189–203. Landau B, Jackendoff R. ‘‘What’’ and ‘‘where’’ in spatial language and spatial cognition. Behavioral and Brain Sciences 1993; 16: 217–66. Levin B. English verb classes and alternations. Chicago: University of Chicago Press, 1993. Lindstromberg S. English prepositions explained. Amsterdam: John Benjamins, 1998. Martin A, Chao LL. Semantic memory and the brain: Structure and processes. Current Opinion in Neurobiology 2001; 11: 194–201. Martin A, Haxby JV, Lalonde FM, Wiggs CL, Ungerleider LG. Discrete cortical regions associated with knowledge of color and knowledge of action. Science 1995; 270: 102–5. McCarthy R, Warrington E. Category specificity in an agrammatic patient: The relative impairment of verb retrieval and comprehension. Neuropsychologia 1985; 23: 709–27. Munnich E, Landau B, Dosher BA. Spatial language and spatial representation: A cross-linguistic comparison. Cognition 2001; 81: 171–207. Newman J, editor. The linguistics of giving. Amsterdam: John Benjamins, 1997. Newman J, editor. The linguistics of sitting, standing, and lying. Amsterdam: John Benjamins, 2002. Papafragou A, Massey C, Gleitman L. Shake, rattle, ‘n’ roll: The representation of motion in language and cognition. Cognition 2002; 84: 189–219. Pawley A. Encoding events in Kalam and English: Different logics for reporting experience. In: Tomlin R, editor. Coherence and grounding in discourse. Amsterdam: John Benjamins, 1993: 329–60. Poldrack RA, Wagner AD, Prull MW, Desmond JE, Glover GH, Gabrieli JDE. Functional specialization for semantic and phonological processing in the left inferior frontal cortex. NeuroImage 1999; 10: 15–35. Pulvermuller F, Hummel F, Harle M. Walking or talking? Behavioral and neurophysiological correlates of action verb processing. Brain and Language 2001; 78: 143–68. Double dissociation between verbs and prepositions Rappaport Hovav M, Levin B. Building verb meanings. In: Butt M, Geuder W, editors. The projection of arguments: Lexical and compositional factors. Stanford: CSLI, 1998: 97–134. Rizzolatti G, Fogassi L, Gallese V. Neurophysiological mechanisms underlying the understanding and imitation of action. Nature Reviews (Neuroscience) 2001; 2: 661–70. Sandra D, Rice S. Network analyses of prepositional meaning: Mirroring whose mind-the linguist’s or the language user’s? Cognitive Linguistics 1995; 6: 89–130. Senior C, Barnes J, Giampietro V, Simmons A, Bullmore ET, Brammer M, Davis AS. The functional neuroanatomy of implicit-motion perception or ‘‘representational momentum’’. Current Biology 2000; 10: 16–22. Shapiro KA, Pascual-Leone A, Mottaghy FM, Gangitano M, Caramazza A. Grammatical distinctions in the left frontal cortex. Journal of Cognitive Neuroscience 2001; 13: 713–20. Slobin DI. The many ways to search for a frog: Linguistic typology and the expression of motion events. In: Stromqvist S, Verhoeven L, editors. Relating events in narrative: Typological and contextual perspectives. Mahwah: Lawrence Erlbaum Associates, in press. Stamenov MI, Gallese V, editors. Mirror neurons and the evolution of brain and language. Amsterdam: John Benjamins, 2002. Talmy L. Lexicalization patterns: Semantic structure in lexical forms. In: Shopen T, editor. Language typology and syntactic description. Cambridge: Cambridge University Press, 1985: 57–149. Tesak J, Hummer P. A note on prepositions. Brain and Language 1994; 46: 463–8. Thompson-Schill SL, D’Esposito M, Kan IP. Effects of repetition and competition on activity in left prefrontal cortex during word generation. Neuron 1999; 23: 513–22. Tranel D. The Iowa-Benton school of neuropsychological assessment. In: Grant I, Adams K, editors. Neuropsychological assessment of neuropsychiatric disorders, 3rd edn. New York: Oxford University Press, 1996: 81–101. Tranel D, Adolphs R, Damasio H, Damasio AR. A neural basis for the retrieval of words for actions. Cognitive Neuropsychology 2001; 18: 655–70. Tranel D, Kemmerer D. Neuroanatomical correlates of English spatial prepositions as revealed by 3D lesion mapping. Manuscript under review. Tranel D, Kemmerer D, Adolphs R, Damasio H, Damasio AR. Neural correlates of conceptual knowledge for actions. Cognitive Neuropsychology, in press. Tranel D, Logan C, Frank R, Damasio AR. Explaining category-related effects in the retrieval of conceptual and lexical knowledge for concrete entities: Operationalization and analysis of factors. Neuropsychologia 1997; 35: 1329–39. Ungerleider LG, Mishkin M. Two cortical visual systems. In: Ingle DJ, Goodale MA, Mansfield RJW, editors. Analysis of visual behavior. Cambridge, MA: MIT Press, 1982: 549–86. Van Valin RD, LaPolla R. Syntax: Structure, meaning and function. Cambridge: Cambridge University Press, 1997. Warburton E, Wise RJS, Price CJ, Weiller C, Hadar U, Ramsay S, Frackowiak RSJ. Noun and verb retrieval by normal subjects: Studies with PET. Brain 1996; 119: 159–79. Wienold G. Lexical and conceptual structures in expressions for movement and space: With reference to Japanese, Korean, Thai, and Indonesian as compared to English and German. In: Egli U, Pause PE, Schwartze C, Stechow A, Wienold G, editors. Lexical knowledge in the organization of language. Amsterdam: Benjamins, 1995: 301–40. Wilson S. Coverbs and complex predicates in Wagiman. Stanford: CSLI, 1999. Wise R, Chollet F, Hader U, Friston K, Hoffner E, Frackowiak R. (1991). Distribution of cortical neural networks involved in word comprehension and word retrieval. Brain 1991; 114: 1803–17. Received on 9 December, 2002; resubmitted on 13 March, 2003; accepted on 24 March, 2003 435 A double dissociation between the meanings of action verbs and locative prepositions David Kemmerer and Daniel Tranel Abstract We describe two patients who manifested opposite patterns of performance on test batteries that evaluated production, comprehension, and semantic analysis of action verbs on the one hand (e.g., smile, wave, run) and locative prepositions on the other (e.g., in, on, over). JP failed all of the verb tests but passed all of the preposition tests, suggesting impaired knowledge of the meanings of action verbs but intact knowledge of the meanings of locative prepositions. In contrast, RR exhibited the reverse dissociation: he passed many of the verb tests but failed all of the preposition tests, suggesting mostly intact knowledge of the meanings of action verbs but impaired knowledge of the meanings of locative prepositions. This behavioral double dissociation reflects the fact that the two categories of words differ along several conceptual parameters. To a large extent, the patients exhibited a neuroanatomical double dissociation as well, since JP’s lesion is predominantly in the left frontal operculum whereas RR’s is predominantly in the left inferior parietal lobe and the posterior superior temporal region. This constitutes preliminary evidence that the meanings of action verbs and locative prepositions are represented by partially independent neural networks in the brain. Journal Neurocase 2003; 9: 421–435 Neurocase Reference Number 564/02 Primary diagnosis of interest Semantic impairments Author’s designation of case JP, RR Key theoretical issue * Neural and cognitive organization of different categories of word meanings Key words: semantic structure; semantic memory; conceptual knowledge; category specificity; verbs; prepositions Scan, EEG and related measures MRI Standardized assessment Western Adult Intelligence Scale (WAIS)-III, Benton Visual Retention Test (BVRT), Multilingual Aphasia Examination, Boston Diagnostic Aphasia Examination (BDAE). facial discrimination, judgment of line orientation, Rey-Osterrieth Complex Figure, three-dimensional block construction, visual object recognition and naming Other assessment Naming, matching, odd one out, and verification tests for verbs and prepositions Lesion location * JP: left frontal (pars opercularis, pars triangularis, pars orbitalis, and middle premotor region, along with the white matter underlying these cortices); left anterior insula; white matter beneath left postcentral gyrus and inferior parietal lobule * RR: cortex and white matter of left supramarginal, angular, and posterior superior temporal gyri; pars opercularis of left frontal lobe; white matter of left temporal pole Lesion type Cerebrovascular accident (stroke) Language English