- Email: [email protected]
Two routes to naming: A case study
Pergamon
0028 3932(95)00166 2
Neuropsycholoyia, Vol. 34, No. 8, pp. 809 817, 1996 Copyright ~" 1996 Elsevier Science Ltd. All rights reserved Printed in Great Britain 0028 3932/96 $15.00+0.00
Two routes to naming: A case study LILIANNE MANNING and ELIZABETH K. WARRINGTON* The National Hospital for Neurology and Neurosurgery, Queen Square, London WC1N 3BG, U.K. (Received 5 June 1995: accepted 3 November 1995)
Abstract--We report the case of an aphasic patient, KP, whose spontaneous speech and ability to communicate verbally were
severely reduced. His comprehension of written words was preserved. Naming was relatively preserved for objects but impaired for actions. We designed experiments to investigate his verb processing and sentence completion. KP's knowledge of verb syntax was impaired in contrast to his relatively preserved knowledge of verb semantics. His ability to retrieve a noun on visual confrontation was significantly better than his ability to retrieve the same pool of nouns through propositional language. We interpret the data in terms of two 'routes' to noun-retrieval, a nominal route and a propositional speech route. Furthermore, we suggest that the dissociation between verb syntax and verb semantics can also be accounted for within this framework. Copyright ~ 1996 Elsevier Science Ltd. Key Words: anomia; aphasia; word retrieval.
"perseverative aphasia" and "dynamic aphasia". In the former syndrome the patient is able to repeat only single words. Dynamic aphasic patients do not have perseverations and, consequently, repetition of sentences is accurate but, here also, spontaneous speech is sparse even to the point of lacking content words. As in the original descriptions of transcortical motor aphasia, these patients' naming is accurate and in marked contrast to their impairment of spontaneous speech. An additional feature observed in some anomic patients is the evidence of selective impairments in the processing of certain classes or categories of words. Dissociations between semantic classes such as animate versus inanimate categories [40, 41] as well as between grammatical classes such as nouns versus verbs have been observed [22, 24, 26, 45]. Naming abilities are usually assessed by means of picture naming tasks. However, the context in which the target is to be produced can affect the level of performance. Williams and Canter [43, 44] demonstrated that Broca's aphasic patients were able to retrieve a noun significantly better on a confrontation naming task than on a picture description task. By contrast, their Wernicke's aphasia patients performed significantly worse on the confrontation naming task than on a picture description task. A further contextual factor that affects the level of naming performance is whether the target is produced as an isolated word, typically naming to visual confrontation, or within a sentential context. Zingeser and Berndt [45] reported that their patient, HY, produced
Introduction
Naming difficulties are observed in all the major aphasic syndromes. In some patients, such difficulties occur in the context of otherwise fluent speech. The opposite pattern of performance, relatively well-preserved naming abilities in the context of markedly non-fluent spontaneous speech was first described by Lichtheim in 1885 [17]. He interpreted this syndrome as the result of disruptions between concept centres and the centres for motor images of words. Lichtheim termed this type of language dysfunction "traumatic aphasia" (case II; p. 447). Wernicke [42] accepted Lichtheim's description of the syndrome and re-termed it as transcortical motor aphasia (TMA) [8]. T M A is characterised by a disproportionately reduced spontaneous speech, compared with verbal comprehension, retained ability to repeat words and sentences, name objects and read aloud. The first published case in which naming isolated items was intact, whereas composition of sentences from given words was defective, was reported by Heilbronner [12; cited by De Bleser, 7]. Perseverations are not uncommon in TMA; when present, they contaminate written, pointing and spoken responses. Thus, Luria and Hutton [20] distinguished two forms of transcortical motor aphasia, *Address for correspondence: Department of Neuropsychology, The National Hospital for Neurology and Neurosurgery, Queen Square, London WC1N 3BG, U.K.; e-mail: [email protected] f.ac. uk 809
810
L. Manning and E. K. Warrington/Routes to naming
nouns significantly better when he was given a sentence frame than when he was asked to name objects to visual confrontation. Similarly, Breen and Warrington's [4] severely anomic patient, NOR, was able to retrieve nouns significantly better within a sentence context than on visual confrontation. Both patients, HY and NOR, presented with a grammatical class dissociation: they named verbs significantly better than nouns. The converse pattern of performance, very poor action naming together with impaired sentence generation, is characteristically shown by agrammatic patients. Some studies have focused on the dissociation of syntactic from semantic processes involved in language reception and production [34]. Further investigations have focused on patients whose productive syntactic deficits selectively affect the constructional aspect of sentence generation, sparing the morphosyntactic processing [1, 28]. In this paper we report a patient, KP, whose spontaneous speech was severely non-fluent and who was unable to generate sentences. This was in contrast to his naming abilities which appeared to be relatively well preserved. Additionally, he presented with a grammatical-class dissociation: his ability to name objects was significantly better than his ability to name actions. Our aim therefore, was to document the scope of KP's impaired verb processing and to investigate to what extent this impairment contributed to his non-fluent aphasia.
Case report KP (d.o.b. 20.12.27), a right-handed retired computer operator was referred to the National Hospital in 1992 for investigations of weakness in his right hand and forgetfulness. On examination there was a mild right-sided ataxia of the cerebellar type, diplopia and a slurring of speech. The only other finding of note was his cognitive decline which affected for the most part his language functions (see below). An MRI scan (27.2.92) showed diffuse high signal lesions predominantly adjacent to the ventricles. Over the next 2 years his neurological status deteriorated and his memory and language skills became more impaired. A dementing illness due to diffuse cerebrovascular disease was diagnosed.
Psychological test.findings KP was first tested on the Wechsler Adult Intelligence Scale-Revised (WAIS-R) in January 1992 when he obtained a verbal IQ of 91 (pro-rated from four subtests) and a performance 1Q of 117 (pro-rated from three subtest scores). On this first assessment, his visual recognition memory was average (42/50) and his verbal recognition memory was low average (39/50) [37]. KP was re-referred to the department on 11 January 1994. Comprehensive formal assessment was no longer possible because of KP's difficulties in grasping instructions and in following test
demands. On this occasion, he was given selective subtests of the WAIS-R. He obtained a low average score in the digit span (six forwards and three backwards). His score on one nonverbal subtest (block design) was superior as it had been previously. His digit span was tested again in September 1994. On this occasion, his span forwards had deteriorated to two digits (he was not tested on digit span backwards). It was observed that identical performance was obtained with other stimuli: thus his span for both letters and words was also two. On a short version of the Verbal Recognition Memory Test he scored below the 5th percentile (17/25) [38]. By contrast, visual recognition memory on the standard, more stringent test was again average (41/50). He was only able to name 4/12 famous faces, however, he recognised a further five stimuli. Language assessment. His spontaneous speech was non-fluent and so sparse that his ability to communicate verbally was severely reduced. His speech was very mildly dysarthric but no phonemic paraphasias were observed. A single word repetition task was presented. The list consisted of one-, two- and three-syllable words; there were 90 words in each of two frequency bands: high frequency (A or AA) and low frequency (<10). KP repeated 140/180 words correctly. He had significantly more difficulty in repeating the low frequency words (63/90 correct) than the high frequency words (77/90 correct, ;(2 = 5.4, P < 0.25). His naming ability was strongly affected by the frequency of the items to be named. He named the 19/30 more common items of the Oldfield test [29]. On a further naming test, consisting of 50 high frequency items, 10 of each of five categories (objects, countries, animals, body parts and colours), performance was fairly satisfactory, he scored 45/50 correct (four errors were made on the animal category). By contrast, naming 30 high frequency verbs was markedly poor (3/30 correct) [22]. KP attempted a sentence completion task. A written sentence with the last word, a high probability terminal noun omitted was presented (e.g., "He sent the letter without a ---"). KP was required to read the sentence and either say aloud or write the missing word. His performance was poor, he was able to retrieve a suitable word (noun) for only 2/5 sentences. His reading abilities were assessed on the National Adult Reading Test (NART) [27] in January 1994 when he obtained an average score (reading IQ equivalent = 90). He was presented with two further sets of words whose spelling to sound correspondences differed in the degree of irregularity. There were 20 "mildly irregular" words (e.g. shovel, castle, comb) and 20 "very irregular" words (e.g. plough, doughnut, parachute). KP scored 16/20 correct on both sets. On a further list of 50 closedclass words (pronouns, adjectives and conjunctions) he scored 43/50 correct. Comprehension at the single word level was assessed on a stringent written two-choice synonym test for concrete and abstract words [39]. He obtained an average
L. Manning and E. K. Warrington/Routes to naming score on both versions (concrete 18/25; abstract 20/25 correct). In view of his very poor span and of his fairly competent reading abilities, all verbal comprehension tests which involved more than one item were presented in the written modality. Five tests were administered to assess his comprehension at the sentence level. All except one test were presented in the written sentence-picture matching modality; the remaining test involved only written-sentences: (a) Written comprehension of locative relations [15]. The aim of this test is to assess comprehension of locative prepositions. The test stimuli consist of 24 four-choice sets of pictures and 24 written sentences which comprise only three items, two referents and their spatial relation (e.g., "Box over bucket"). No verbs are included in the phrases. KP scored 21/24 correct which is at the same level as the mean for the control subjects (mean = 21.84). (b) Schwartz et al. 's reversible sentences [34]. This task assesses word order comprehension in two classes of constructions: (i) 23 active sentences using twochoice pictures. The subject and object of each picture are "stick-men" cartoons with round or square heads (e.g., "The circle paints the square"); (ii) 23 locative sentences using same geometrical shapes, circle and square (e.g., "The square is in the circle"). The sentences were presented in the written modality. KP performed these tasks competently; he scored 23/23 correct on the active sentences and 21/23 correct on the locative sentences. (c) Sentence-picture matching: written version [15]. The aim of this test is to assess the comprehension of written sentences involving different grammatical constructions. The test consists of 60 sentences and 60 sets of three pictures. The sentences are divided into four main types of constructions: reversible (e.g., "The dog is approaching the girl"); nonreversible (e.g., "The cat is licking the man"); gapped (e.g., "The girl is considering where to go") and converse relations (e.g., "The man is offering the money"). KP's performance was impaired for reversible sentences for which he obtained a score of 13 correct (controls" mean: 19.15). He performed creditably on two further categories (gapped sentences, KP: 12; controls: 15.3. Converse relations, KP: 5; controls: 6.85) and competently on the remaining class (non-reversible sentences, KP: 16; controls: 15.48). (d) Testjor reception of grammar ( TROG) [2]. This test assesses comprehension of grammatical contrast. The test stimuli consist of 80 sentences graded in grammatical complexity and 80 sets of four pictures. In the standard version of the test, the subject is asked to select the picture that corresponds to a spoken sentence'. KP was asked to select the picture that corresponded to a written
811
sentence. His performance was weak, he scored 63/80 correct (12/17 errors were made on locative sentences, e.g., "The pencil is on the box"). (e) The word-class comprehension test [6]. The aim of this test is to assess comprehension of grammatical word-classes within a written sentence context. The test consists of 105 fairly complex sentences in which, for the target word, there are three alternatives. Twenty-four sentences test prepositions (e.g., "The lady-in-waiting off/at/to the proud princess is dismissed by the queen"); 15 sentences test adjectives (e.g., "The serious/coloured/blunt lights at the Palladium were operated by the assistant producer"); 33 sentences test nouns (e.g., "The poultry farmers were plucking the feathers off the squawking mice/leopards/geese"). The remaining 33 sentences test verb tense (e.g., "The boy hits/was hit/were hit on the head by the falling stone"). KP had to select one of the three alternatives given in each sentence. Our patient's performance with the nouns was satisfactory (97% correct); his performance was adequate when he was asked to select adjectives (72% correct) but it was poor for verbs (57% correct) and at chance level for prepositions (33% correct). Overall, KP's comprehension of written sentences was fairly satisfactory with two exceptions. His comprehension of verb tense was impaired as was his comprehension of prepositions hut only in the context of complex sentences. A series of experiments were designed in order to investigate KP's verbal production and syntactic processing in more detail.
Experimentalinvestigation Test 1. Sentence production It was observed qualitatively that our patient had virtually no spontaneous speech. Our aim in this section therefore was to document more precisely KP's difficulties in producing sentences. Two tests were devised which assessed his ability to generate sentences. Sentence generation 1. The aim of this task was to investigate whether KP was able to generate sentences in a condition in which he could "use" a sentence structure given by the examiner to facilitate his own sentence production. Eighty sets of sentences together with pictures from the T R O G (see above) were presented to KP. The test was completed in three stages. First, the sentence structure was demonstrated by presenting a picture together with the appropriate sentence (e.g., "'The flower is longer than the pencil"). Second, KP was required to construct a sentence to describe a second picture using the same format given the prompt of the subject noun (e.g., "The k n i f e . . . " ) . Third, KP was required to con-
812
L. Manning and E. K. Warrington/Routes to naming
struct a sentence using the same format to describe a third picture without the prompt of the subject noun. Results. KP was unable to produce a single grammatically correct sentence, regardless of the presence or absence of a prompt. Whenever he attempted to generate a sentence his production was limited to naming parts of the picture (e.g., "The boy, the elephant is the b o y - push"; "the brown cat--the white cat"). Sentence generation 2. Our aim in this test was to investigate whether our patient was able to form sentences in a condition in which no word-retrieval was necessary, using a procedure similar to that reported by Von Stocker [36]. The test stimuli consist of 20 six- to nine-word written sentences from the T R O G . Each individual sentence was displayed in front of KP with the individual words in random order together with the corresponding picture. KP was required to re-arrange the words in order to obtain the sentence which described the picture. Three practice trials were given and no time limit was set. Results. KP was able to re-arrange 6/20 sentences correctly (e.g., correct sentence: "The boy is sitting but not eating"; e.g., an incorrect sentence: "The girl chases that is big the dog"). Comment. It has been demonstrated that KP's difficulties in producing sentences are not exclusively due to word-retrieval problems since his score did not improve significantly in the second condition in which wordretrieval was not required.
Test 2. Naming, reading and spelling objects and actions It was observed that our patient's action naming was markedly poor in comparison with his object naming. Our aim in this experiment therefore was to assess in more detail KP's ability to process verbs and nouns. The test stimuli consists of 75 objects/nouns (e.g., key, bridge, towel, zip) and 75 actions/verbs (e.g., read, leave, pat, skate) matched in frequency [30]. KP was required to name, read and spell each test item. The stimuli used in the naming test consisted of very clear and unambiguous line drawings selected from several published word-picture matching tasks. Naming and spelling were tested in an ABBA design and reading was tested in a subsequent session. Results. The percentage correct for verbs and nouns for each condition is presented in Table 1. The results
Table 1. KP's performance: Naming, reading and spelling the same pool of verbs and the same pool of nouns. Percentage correct
Naming Reading Spelling
n
Verbs %
Nouns %
75 75 75
19 99 83
75 99 83
show that his ability to name action/verb pictures was significantly worse than his ability to name object/noun pictures (Z2 = 54, P < 0.0001). His reading was set at ceiling for both nouns and verbs and there was a low but identical spelling error rate for nouns and verbs. Comment. It has been demonstrated that KP is able to read and spell nouns and verbs at a similarly competent level. However, there was a clear-cut word-class effect for naming.
Test 3. Action and object picture-word matching Our aim in this test was to investigate KP's comprehension of verbs and nouns using the word-picture matching test devised by McCarthy and Warrington [22]. The stimuli consist of 40 pairs of unambiguous line drawings, half of them are related pairs of objects (e.g., barbecue/casserole) and half are related pairs of actions (e.g., painting/drawing). The examiner spoke one of the items shown on each picture and KP had to point to the corresponding item. In order to probe both stimuli the picture-pairs were presented twice. Results. KP's performance was satisfactory for both, objects (97% correct) and actions (90% correct). Comment. The fact that KP's comprehension of semantically related verbs is preserved implies that his deficient action verb naming (experiment 2) was not due to degraded semantic knowledge of verbs.
Test 4. Concrete/abstract verb comprehension It was observed that KP was well able to perform the written version of a stringent test of verbal comprehension (Concrete/Abstract Comprehension Test; see above) including those few instances of verbs (e.g., hoist is it lift or carry?; sever is it cut or prune?; cogitate is it meditate or hesitate?). A concrete/abstract verb test was devised to confirm that his verb comprehension was equally intact. First, a pilot study was carried out to establish the concreteness rating of a pool of verbs. Six normal control subjects rated these verbs on a 7-point scale. This list of verbs was presented to four further controls in a twochoice format: a higher frequency synonym and a distractor item. On the basis of their results two lists of 25 verbs, one with a high concrete rating (e.g., confine does it mean imprison or end?) and one with a high abstract rating (e.g., acquiesce does it mean acquire or agree?) were selected. This written concrete and abstract 50 verb synonym test was presented to KP. Three controls matched in their reading IQ equivalent to KP were also tested. Results. The percentage correct for KP and the control
L. Manning and E. K. Warrington/Routes to naming subjects for the concrete and abstract verb comprehension are given in Table 2. The critical finding is that KP's performance on the abstract set was identical to that of the control subjects. Comment. KP's relatively preserved verb meaning, as documented in the concrete/abstract very synonym test, indicates that his impaired action naming cannot be attributed to impaired comprehension of verbs. This finding corroborates the results of the action word-picture matching test. Furthermore, his very satisfactory comprehension of abstract verbs indicates that the noun verb discrepancy cannot be accounted for in terms of a concrete/abstract dissociation.
Test 5. Written sentence-pieture matching test." verb tenses We have reported (see above) that KP had selective difficulties with verbs on a task where he was asked to choose the correct alternative within a sentence frame (the word-class comprehension test). We have also documented KP's relatively preserved ability to discriminate between semantically related verbs and to comprehend. The aim in this test is to investigate whether the disparity of results between impaired use of verbs and relatively preserved verb semantics could be due to a defective processing of verb tense. The test stimuli consist of 20 pairs of pictures, drawn from the Parisi and Pizzamiglio [31] set. Each pair shows unambiguously two tenses of the same action performed by the same actor (the actions were performed in the present, present simple, present progressive, past, past perfect and future tenses). A short sentence was written below the pairs of pictures (e.g., "The boy will draw"). KP was requested to match the written sentence to the appropriate picture. Results. KP's performance was at chance level: he scored 10/20 correct. For example, a picture of "the girl will drink" was matched with the picture of '~a girl drinking". ~'The boy has been running" was matched with "the boy is running". Comment. We have demonstrated that KP was no longer able to process verb tense; his performance was markedly impaired when he had to select a given tense from only two possibilities. This finding corroborates
Table 2. KP's and three controls' performance on the synonym concrete and abstract word comprehension test
Concrete Abstract
KP
Controls n -- 3 (mean)
19/25 20/25
22/25 20/25
813
KP's difficulties with verb tense as reported in the wordclass comprehension test.
Test 6. Sentence completion and con/rontation naming Breen and Warrington [4] have recently demonstrated that their anomic patient (NOR) who was insensitive to phonological and semantic cues improved his object naming significantly when a sentence frame was provided, i.e. he was able to provide the terminal noun which had been omitted in a sentence. Our aim in this experiment was to document KP's sentence completion skills. Clinical observation suggested that his abilities to complete a sentence with a high probability terminal word were very poor (see above). Two tests were devised: Sentence completion 1. Thirty sentences were constructed such that there was a medium or low probability that a particular word would be generated in the terminal position (e.g., ~'He went to the stationery shop to buy me s o m e . . 2' has a medium probability; "I was looking for t h e . . . " has a low probability). KP was required to read each sentence and generate a missing last word. Any plausible completion was accepted as correct. Results. The percentage correct for each level of probability was as follows: • Medium probability sentences: 46% correct; • Low probability sentences: 13% correct. Although KP's score was significantly better with the medium probability sentences than the low probability sentences 0~2 = 3.99, P < 0.05), his performance on this task was still quite impaired. Sentence completion 2. We have reported in test 2 (see above) KP's fairly adequate naming performance for objects (56/75 correct). In the present test, we re-presented the same 56 objects and the patient was required to name them. Additionally, for each object name there was a corresponding sentence frame with the target terminal noun omitted. In every case, the object name formed a high probability completion of the sentence (e.g., "The archer picked up his bow and fired a n . . . " ) . The sentences were written and no time limit was set. KP was asked to generate the terminal word. The two tasks, naming on visual confrontation and sentence completion, were presented in a ABBA design. Results. The few plausible responses which did not correspond to our target word were accepted as correct. KP's performance on the sentence completion task was poorer (34% correct) than his performance on visual confrontation naming (89% correct). There was a highly significant difference in retrieving the object's name in the two conditions (;~2 = 47, P < 0.0001). Comment. We have demonstrated in the first task that KP had great difficulty in completing low probability sentences and his performance showed only moderate improvement with medium probability sentences. More importantly, we have demonstrated in the second task a clear dissociation between noun-retrieval by visual confrontation which appears relatively well preserved and
814
L. Manning and E. K. Warrington/Routes to naming
noun-retrieval through sentence completion which is markedly impaired.
Discussion
We have presented a patient in whom the following findings were documented: KP had markedly defective sentence generation. He showed a clear-cut grammatical class dissociation with action naming significantly worse than object naming. Although he was able to name objects on visual confrontation, he was unable to retrieve those very same object names on a sentence completion task in which he was provided with a high probability frame to elicit the terminal noun. It was also demonstrated that he was able to comprehend the meaning of verbs on a word-picture matching task and on a more stringent synonym concrete/abstract verb comprehension test. However, his preserved knowledge of the semantics of verbs contrasted with his inability to process the syntax of verbs, as it was demonstrated by his very impaired performance on verb tense tasks. KP's sentence comprehension on the written modality was generally adequate, except for the following grammatical constructions: his performance on prepositional verb sentences was weak in complex constructions. He was able to comprehend reversible sentences in constructions in which animacy/saliency was neutralised. Sentences in which verb tense was crucial were poorly understood. First, we discuss our patient's deficient sentence generation. Second, we examine his verb processing in relation to both his impaired sentence generation and his inability to use high probability sentence frames to produce nouns which he was able to produce on visual confrontation. The patient's inability to generate sentences occurred within the context of a relatively well-preserved wordretrieval ability on visual confrontation tests. KP's failure to generate sentences does not seem to be due to the characteristics generally attributed to "dynamic aphasia" either in Luria's [19, 21] or Costello and Warrington's [5] account of this syndrome. The former authors stated that dynamic aphasia is due to a marked disturbance in spontaneous expanded speech, all other aspects of language remaining intact. Costello and Warrington argued that it is the incapacity to organise or plan thoughts which causes the symptoms of dynamic aphasia, rather than the inability to implement that planning. KP did indeed seem unable to organise or plan thoughts. However, his inability to process verb tenses together with his difficulties in complex locative prepositional sentences and some reversible sentences discourage any attempt to account for his sentence generation impairment in terms of dynamic aphasia. By the same token, his impaired syntactic processing leads us to consider some evidence on grammatical issues. Grammatical difficulties in sentence production were classified in two different types of agrammatism, m o r -
phological and syntactic, by Tissot et al. [35]. The morphological agrammatic patients would have problems with function words and word endings. KP presented with both classes or errors. The syntactic agrammatic patients would have difficulties with the word order and the retrieval of verbs. KP had difficulties with word order when he was asked to organise written words into sentences. However, his speech was so sparse that it was hardly possible to detect spontaneous word order difficulties. He also had difficulties in choosing verb tenses. There are two main accounts of sentence generation, Garrett's model [9, 10] and Bock's model [3]. We will comment briefly on these two models and consider which, if either, captures KP's deficit in sentence production. Garrett's model postulates a division between lexical retrieval and the generation of a functional argument structure, both of which belong in the "message-level representation". In cases in which the lexical retrieval ability is preserved but the possibility of creating a functional predicate/argument structure is impaired, the patients would be left with a collection of content words but no predicate/argument structure in which to insert them. However, KP's inability to generate sentences cannot be explained by this model. Although he was able to name single objects, he had a grammatical class dissociation. KP's impaired action naming does not seem to be readily explained by a theoretically preserved lexical retrieval function which makes no grammatical-class specifications. Bock [3] proposed that verb retrieval is crucial in sentence generation because verbs carry the information necessary to assign thematic roles. A verb's specification of functional relations is responsible for the order in which the semantically specified and grammatically tagged words will be linked together. In Bock's sentence generation model, input from several levels may contribute to the process of generating sentences, and at each stage verbs play a central role in subsequent aspects of sentence production. KP's defective verb tense processing and action naming, therefore, must be implicated in his inability to generate sentences. Although the terms noun and verb refer to syntactic classes, they elicit meaning differences. Evidence has emerged that there are psychological distinctions in the processing and representation of nominal and verb meaning [33]. Besides the well-known dissimilarities between verbs and nouns in terms of the former being less memorable, harder to interpret and possessing a wider range of meanings across languages, the functional differences are more relevant in our attempt to interpret our patient's performances. KP's pattern of performance in relation to verbs suggested a dissociation between, on the one hand, the use of rules that command the syntactic aspects of verbs, more specifically, verb tenses and, on the other hand, the comprehension of verb meaning including some fairly infrequent instances of verbs. This finding, with regard to his impaired sentence generation would imply that
L. Manning and E. K. Warrington/Routes to naming comprehension of verbs (relatively preserved in KP) is not sufficient to generate sentences and that adequate verb conjugation is crucially necessary. This suggestion is parallel to that advanced by several authors who studied other syntactic constructions. For example, Nespoulous et al. [28] studied a patient whose difficulties were exclusively observed in the production of sentences due to deficits at the level of "use" of morphology rather than defective syntactic knowledge. A further subtle dissociation was studied by Linebarger et al. [18]. They came to the conclusion that there may be a dissociation between the ability to perceive syntax (the patients detected over 90% of grammatically wrong constructions) and the ability to comprehend syntax (they scored at chance level on picture-pointing tests in which word order was crucial). It is perhaps not surprising then, that patients whose semantic and syntactic verb processing is spared have a preserved performance on sentence generation tasks despite anomic difficulties. What appears to be more interesting is the finding that verb syntax is necessary in order to make use of sentence frames in naming tasks. Noun production in sentence completion tasks seems to be the result of verbal activations that are not present in other forms of verbal prompting such as, for example, definitions nor in naming from visual confrontation. Zingeser and Berndt [45] described a patient (HY) with marked naming difficulties whose performance on sentence completion tasks was virtually error free (95% correct on high probability sentence frames versus 70% correct on naming on visual confrontation). HY's verb production was far superior to noun production in single word naming (62.5% naming action pictures in contrast to 30-35% naming objects). A further case report with very similar findings is the study by Breen and Warrington [4]. Their patient NOR's naming skills were markedly impaired. Phonological and semantic cues were ineffective in helping him retrieve a given noun. In contrast, a sentence frame, independent of the level of probability to elicit a target had a significant facilitatory effect. Thus, the patient named 76% of pictures when presented together with an incomplete sentence versus 29% correct on visual confrontation tasks without sentence frames. NOR's grammatical-class dissociation was similarly clear-cut: he could name only 15% of the coloured pictures of objects but 68% of pictured actions. Zingeser and Berndt's and Breen and Warrington's data suggest an integral association between preserved action naming and the ability to use sentential frames in order to retrieve nouns which are not retrieved in a confrontational naming task. KP presented with the converse pattern of performance: deficient action naming and inability to use sentence frames to retrieve nouns which he was able to retrieve in an object naming task. We would like to interpret this case following the argument put forward by Breen and Warrington [4] who argued for there being two routes to phonology, the nominal and the propositional routes.
815
It is interesting to recall that as early as 1876, Hughlings Jackson [13] distinguished between propositional speech and automatic speech. This distinction has been elaborated and refined by Head [11], Pick [32] and Kleist [16] who in 1934 used the already well-established distinction between "nominalising" speech and "propositionising" speech. In linguistics, the term ~'nominalising" corresponds to the paradigmatic speech which is used in naming and the term "propositionising" corresponds to the syntagmatic speech which refers to the act of producing sentences in which words are modified by the context [14]. Similarly, Luria [19] differentiated between nominative and propositional language. Breen and Warrington emphasised the possibility that noun-retrieval is the result of two different procedures: noun-retrieval in propositional speech and in nominal contexts. They interpreted their case, NOR, who was very impaired in confrontation naming tasks but obtained a significant benefit from sentence contexts, in terms of the operation of an on-line, integrative language processor. Characteristically, this processor is thought to involve a shortterm memory system and to operate both at a level of processing complexity beyond single words, and in an anticipatory mode. In other words, the language processor constructs propositional speech [23, 25], This process would constitute one of the two routes to phonology, the other route being one which involves the procedures of naming to confrontation. Following their hypothesis, KP's relatively preserved performance on naming tasks would indicate a preserved nominal route whereas his marked inability to generate sentences and, in particular, his incapacity to benefit from sentence frames to name an object would be consistent with such a damaged integrative language processor, the other route to phonology. Finally, it has been documented that: (i) verbs appear to be crucial in sentence generation; (ii) the provision of sentence frames had highly beneficial facilitatory effects in the production of a content word in patients whose action naming was better preserved than their object naming; (iii) our patient presented with impaired action naming and defective verb syntax in the context of relatively preserved verb semantics: and (iv) KP was unable to benefit from the provision of sentence frames in producing a noun. From the perspective of the notion of two routes to naming, these data suggest that verb syntax and/or action naming are crucial for "'using" sentences as verbal prompters, i.e. the propositional route probably hinges on intact or nearly-perfect verb processing. The obvious next stage is to study the effects on naming via propositional speech of selectively impaired semantics of verbs, syntax of verbs and action naming. In summary, our patient presented with a pattern of performance which provides sufficient information to be interpreted in terms of a double dissociation vis-Zt-vis both Zingeser and Berndt's, and Breen and Warrington's patients: KP's defective action naming, on the one hand, contrasted with both HY's and NOR's verb production which was far superior to noun production in single word
816
L. Manning and E. K. Warrington/Routes to naming
naming. Moreover, KP's marked inability to benefit from sentence frames to retrieve a highly predictable terminal noun was the exact opposite to HY's and NOR's performance.
Acknowledgement--We are most grateful to Dr A. J. Lees for permission to study a patient under his care and to report our findings.
References
1. Berndt, R. Symptoms co-occurrence and dissociation in the interpretation of agrammatism, In The Cognitive Neuropsychology of Language, M. Coltheart, G. Sartori and R. Job (Editors), pp. 221 233. Lawrence Erlbaum Associates, London, 1987. 2. Bishop, D. Test for the Reception of Grammar. Medical Research Council. Chapel Press, England, 1990. 3. Bock, K. J. Co-ordinating words and syntax in speech plans. In Progress in the Psychology of Language, A. Ellis (Editor). Lawrence Erlbaum Associates, London, 1987. 4. Breen, K. and Warrington, E. K. A study ofanomia: Evidence for a distinction between nominal and propositional language. Cortex 30, 231-245, 1994. 5. Costello, A. de L. and Warrington, E. K. Dynamic aphasia. The selective impairment of verbal planning. Cortex 25, 103-114, 1989. 6. Coughlan, A. The Word Class Comprehension Test. Unpublished. 7. De Bleser, R. From agrammatism to paragrammatism: German aphasiological traditions and grammatical disturbances. Cognitive Neuropsychology 4, 187-256, 1987. 8. Freud, S. On Aphasia: A Critical Study. (Translated by E. Stengel.) International Universities Press, New York, 1891. 9. Garrett, M. F. Levels of processing in sentence production. In Language Production. Vol. 1. Speech and Talk, B. Butterworth (Editor). Academic Press, London, 1980. 10. Garrett, M. F. Production of speech: Observation from normal and pathological language use. In Normality and Pathology in Cognitive Functions, A. Ellis (Editor). Academic Press, London, 1982. 11. Head, H. Aphasia and Kindred Disorders of Speech. Cambridge University Press, Cambridge, 1926. 12. Heilbronner, K. Uber Agrammatismus und die Storung der inneren Spraghe. Archivfiir Psychiatrie und Nervenkrankheiten 41,653 683. 13. Hughlings Jackson, J. 1876. Reprinted in Selected Writings of John Hughlings Jackson, J. Taylor (Editor). Hodder and Stoughton, London, 1932. 14. Jacobson, R. Studies on child language and aphasia. Janua Linguarum Series Minor 144, Mouton, The Hague, 1971. 15. Kay, J., Lesser, R. and Coltheart, M. PALPA. Lawrence Erlbaum Associates, London, 1992. 16. Kleist, K. Gehirnapathologies. Barth, Leipzig, 1934. 17. Lichtheim, L. On aphasia. Brain 7, 433-484, 1885.
18. Linebarger, M. C., Schwartz, M. F. and Saffran, E. M. Sensitivity to grammatical structure in so-called agrammatic aphasics. Cognition 13, 361-392, 1983. 19. Luria, A. Traumatic Aphasia. Mouton, The Hague, 1970. 20. Luria, A. and Hutton, J. T. A modern assessment of the basic forms of aphasia. Brain and Language 4, 129-151, 1977. 21. Luria, A. and Tsvetkova, L. S. The mechanisms of dynamic aphasia. Foundations of Language 4, 296307, 1968. 22. McCarthy, R. A. and Warrington, E. K. Category specificity in an agrammatic patient: The relative impairment of verb retrieval and comprehension. Neuropsychologia 23, 709-727, 1985. 23. McCarthy, R. A. and Warrington, E. K. Understanding: A function of short term memory? Brain 110, 1565-1578, 1987. 24. Manning, L. and Campbell, R. Optic aphasia with spared action naming. A description and possible loci of impairment. Neuropsychologia 6, 587-593, 1992. 25. Marslen-Wilson, W. D. and Tyler, L. K. The temporal structure of spoken language comprehension. Cognition 6, 1-71, 1980. 26. Miceli, G., Silveri, M. C., Villa, G. and Caramazza, A. On the basis for the agrammatic's difficulty in producing main verbs. Cortex 20, 207-220, 1984. 27. Nelson, H. E. and Willison, J. R. The National Adult Reading Test. The NFER-Nelson Publishing Co., Windsor, England, 1991. 28. Nespoulous, J. L., Dordain, M., Peron, C., Ska, B., Bub, D., Caplan, D., Mehler, J. and Roch Lecours, A. Agrammatism in sentence production without comprehension deficits: Reduced availability of syntactic structures and/or of grammatical morphemes? A case study. Brain and Language 33, 273 295, 1988. 29. Oldfield, R. C. and Wingfield, A. Response latencies in naming objects. Quarterly Journal of Experimental Psychology 17, 273 281, 1965. 30. Orpwood, L. and Warrington, E. K. Word specific impairments in naming and spelling but not reading. Cortex 31,239-265, 1995. 31. Parisi, D. and Pizzamiglio, L. Syntactic comprehension in aphasia. Cortex 6, 204-215, 1970. 32. Pick, A. 1931. Aphasia. Reprinted in translation by Brown J. W. Charles C. Thomas, Springfields, IL, 1973. 33. Reyna, V. F. Understanding verbs: Easy extension, hard comprehension. In Progress in the Psychology of Language, A. Ellis (Editor). Lawrence Erlbaum Associates, London, 1987. 34. Schwartz, M. F., Saffran, E. M. and Marin, O. S. M. The word order problem in agrammatism. I. Comprehension. Brain and Language 10, 249 262, 1980. 35. Tissot, R., Mounin, G. and Lhermitte, F. L'Agrammatism. Dessart, Paris, 1973. 36. Von Stockert, T. R. Recognition of syntactic structure in aphasic patients. Cortex 8, 323-354, 1972. 37. Warrington, E. K. Recognition Memory Test. The NFER-Nelson Publishing Co., Windsor, England, 1984. 38. Warrington, E. K. and Clegg, F. Recognition Memory Test, Shortened Versions. Unpublished.
L. Manning and E. K. Warrington/Routes to naming 39. Warrington, E. K., Jackson, M., Orpwood, L. and McCarthy, R. A. Concrete and Abstract Word Comprehension Test. Unpublished. 40. Warrington, E. K. and McCarthy, R. A. Category access dysphasia. Brain 106, 859-878, 1983. 41. Warrington, E K. and Shallice, T. Category specific semantic impairments. Brain 107, 829 853, 1984. 42. Wernicke, C. Nervenheilkunde. Die neueren Arbeinten uber Aphasie. Fortschritte der Medizin 824~830, 1885.
817
43. Williams, S. and Canter, G. The influence of situational context on naming performance in aphasic syndromes. Brain and Language 6, 63-74, 1982. 44. Williams, S. and Canter, G. Action naming performance in four syndromes of aphasia. Brain and Language 32, 124-136, 1987. 45. Zingeser, L. and Berndt, R. Grammatical class and content effects in a case of pure anomia: Implications for models of lexical processing. Cognitive Neuropsycholo,qy 5, 473-516, 1988.
0028 3932(95)00166 2
Neuropsycholoyia, Vol. 34, No. 8, pp. 809 817, 1996 Copyright ~" 1996 Elsevier Science Ltd. All rights reserved Printed in Great Britain 0028 3932/96 $15.00+0.00
Two routes to naming: A case study LILIANNE MANNING and ELIZABETH K. WARRINGTON* The National Hospital for Neurology and Neurosurgery, Queen Square, London WC1N 3BG, U.K. (Received 5 June 1995: accepted 3 November 1995)
Abstract--We report the case of an aphasic patient, KP, whose spontaneous speech and ability to communicate verbally were
severely reduced. His comprehension of written words was preserved. Naming was relatively preserved for objects but impaired for actions. We designed experiments to investigate his verb processing and sentence completion. KP's knowledge of verb syntax was impaired in contrast to his relatively preserved knowledge of verb semantics. His ability to retrieve a noun on visual confrontation was significantly better than his ability to retrieve the same pool of nouns through propositional language. We interpret the data in terms of two 'routes' to noun-retrieval, a nominal route and a propositional speech route. Furthermore, we suggest that the dissociation between verb syntax and verb semantics can also be accounted for within this framework. Copyright ~ 1996 Elsevier Science Ltd. Key Words: anomia; aphasia; word retrieval.
"perseverative aphasia" and "dynamic aphasia". In the former syndrome the patient is able to repeat only single words. Dynamic aphasic patients do not have perseverations and, consequently, repetition of sentences is accurate but, here also, spontaneous speech is sparse even to the point of lacking content words. As in the original descriptions of transcortical motor aphasia, these patients' naming is accurate and in marked contrast to their impairment of spontaneous speech. An additional feature observed in some anomic patients is the evidence of selective impairments in the processing of certain classes or categories of words. Dissociations between semantic classes such as animate versus inanimate categories [40, 41] as well as between grammatical classes such as nouns versus verbs have been observed [22, 24, 26, 45]. Naming abilities are usually assessed by means of picture naming tasks. However, the context in which the target is to be produced can affect the level of performance. Williams and Canter [43, 44] demonstrated that Broca's aphasic patients were able to retrieve a noun significantly better on a confrontation naming task than on a picture description task. By contrast, their Wernicke's aphasia patients performed significantly worse on the confrontation naming task than on a picture description task. A further contextual factor that affects the level of naming performance is whether the target is produced as an isolated word, typically naming to visual confrontation, or within a sentential context. Zingeser and Berndt [45] reported that their patient, HY, produced
Introduction
Naming difficulties are observed in all the major aphasic syndromes. In some patients, such difficulties occur in the context of otherwise fluent speech. The opposite pattern of performance, relatively well-preserved naming abilities in the context of markedly non-fluent spontaneous speech was first described by Lichtheim in 1885 [17]. He interpreted this syndrome as the result of disruptions between concept centres and the centres for motor images of words. Lichtheim termed this type of language dysfunction "traumatic aphasia" (case II; p. 447). Wernicke [42] accepted Lichtheim's description of the syndrome and re-termed it as transcortical motor aphasia (TMA) [8]. T M A is characterised by a disproportionately reduced spontaneous speech, compared with verbal comprehension, retained ability to repeat words and sentences, name objects and read aloud. The first published case in which naming isolated items was intact, whereas composition of sentences from given words was defective, was reported by Heilbronner [12; cited by De Bleser, 7]. Perseverations are not uncommon in TMA; when present, they contaminate written, pointing and spoken responses. Thus, Luria and Hutton [20] distinguished two forms of transcortical motor aphasia, *Address for correspondence: Department of Neuropsychology, The National Hospital for Neurology and Neurosurgery, Queen Square, London WC1N 3BG, U.K.; e-mail: [email protected] f.ac. uk 809
810
L. Manning and E. K. Warrington/Routes to naming
nouns significantly better when he was given a sentence frame than when he was asked to name objects to visual confrontation. Similarly, Breen and Warrington's [4] severely anomic patient, NOR, was able to retrieve nouns significantly better within a sentence context than on visual confrontation. Both patients, HY and NOR, presented with a grammatical class dissociation: they named verbs significantly better than nouns. The converse pattern of performance, very poor action naming together with impaired sentence generation, is characteristically shown by agrammatic patients. Some studies have focused on the dissociation of syntactic from semantic processes involved in language reception and production [34]. Further investigations have focused on patients whose productive syntactic deficits selectively affect the constructional aspect of sentence generation, sparing the morphosyntactic processing [1, 28]. In this paper we report a patient, KP, whose spontaneous speech was severely non-fluent and who was unable to generate sentences. This was in contrast to his naming abilities which appeared to be relatively well preserved. Additionally, he presented with a grammatical-class dissociation: his ability to name objects was significantly better than his ability to name actions. Our aim therefore, was to document the scope of KP's impaired verb processing and to investigate to what extent this impairment contributed to his non-fluent aphasia.
Case report KP (d.o.b. 20.12.27), a right-handed retired computer operator was referred to the National Hospital in 1992 for investigations of weakness in his right hand and forgetfulness. On examination there was a mild right-sided ataxia of the cerebellar type, diplopia and a slurring of speech. The only other finding of note was his cognitive decline which affected for the most part his language functions (see below). An MRI scan (27.2.92) showed diffuse high signal lesions predominantly adjacent to the ventricles. Over the next 2 years his neurological status deteriorated and his memory and language skills became more impaired. A dementing illness due to diffuse cerebrovascular disease was diagnosed.
Psychological test.findings KP was first tested on the Wechsler Adult Intelligence Scale-Revised (WAIS-R) in January 1992 when he obtained a verbal IQ of 91 (pro-rated from four subtests) and a performance 1Q of 117 (pro-rated from three subtest scores). On this first assessment, his visual recognition memory was average (42/50) and his verbal recognition memory was low average (39/50) [37]. KP was re-referred to the department on 11 January 1994. Comprehensive formal assessment was no longer possible because of KP's difficulties in grasping instructions and in following test
demands. On this occasion, he was given selective subtests of the WAIS-R. He obtained a low average score in the digit span (six forwards and three backwards). His score on one nonverbal subtest (block design) was superior as it had been previously. His digit span was tested again in September 1994. On this occasion, his span forwards had deteriorated to two digits (he was not tested on digit span backwards). It was observed that identical performance was obtained with other stimuli: thus his span for both letters and words was also two. On a short version of the Verbal Recognition Memory Test he scored below the 5th percentile (17/25) [38]. By contrast, visual recognition memory on the standard, more stringent test was again average (41/50). He was only able to name 4/12 famous faces, however, he recognised a further five stimuli. Language assessment. His spontaneous speech was non-fluent and so sparse that his ability to communicate verbally was severely reduced. His speech was very mildly dysarthric but no phonemic paraphasias were observed. A single word repetition task was presented. The list consisted of one-, two- and three-syllable words; there were 90 words in each of two frequency bands: high frequency (A or AA) and low frequency (<10). KP repeated 140/180 words correctly. He had significantly more difficulty in repeating the low frequency words (63/90 correct) than the high frequency words (77/90 correct, ;(2 = 5.4, P < 0.25). His naming ability was strongly affected by the frequency of the items to be named. He named the 19/30 more common items of the Oldfield test [29]. On a further naming test, consisting of 50 high frequency items, 10 of each of five categories (objects, countries, animals, body parts and colours), performance was fairly satisfactory, he scored 45/50 correct (four errors were made on the animal category). By contrast, naming 30 high frequency verbs was markedly poor (3/30 correct) [22]. KP attempted a sentence completion task. A written sentence with the last word, a high probability terminal noun omitted was presented (e.g., "He sent the letter without a ---"). KP was required to read the sentence and either say aloud or write the missing word. His performance was poor, he was able to retrieve a suitable word (noun) for only 2/5 sentences. His reading abilities were assessed on the National Adult Reading Test (NART) [27] in January 1994 when he obtained an average score (reading IQ equivalent = 90). He was presented with two further sets of words whose spelling to sound correspondences differed in the degree of irregularity. There were 20 "mildly irregular" words (e.g. shovel, castle, comb) and 20 "very irregular" words (e.g. plough, doughnut, parachute). KP scored 16/20 correct on both sets. On a further list of 50 closedclass words (pronouns, adjectives and conjunctions) he scored 43/50 correct. Comprehension at the single word level was assessed on a stringent written two-choice synonym test for concrete and abstract words [39]. He obtained an average
L. Manning and E. K. Warrington/Routes to naming score on both versions (concrete 18/25; abstract 20/25 correct). In view of his very poor span and of his fairly competent reading abilities, all verbal comprehension tests which involved more than one item were presented in the written modality. Five tests were administered to assess his comprehension at the sentence level. All except one test were presented in the written sentence-picture matching modality; the remaining test involved only written-sentences: (a) Written comprehension of locative relations [15]. The aim of this test is to assess comprehension of locative prepositions. The test stimuli consist of 24 four-choice sets of pictures and 24 written sentences which comprise only three items, two referents and their spatial relation (e.g., "Box over bucket"). No verbs are included in the phrases. KP scored 21/24 correct which is at the same level as the mean for the control subjects (mean = 21.84). (b) Schwartz et al. 's reversible sentences [34]. This task assesses word order comprehension in two classes of constructions: (i) 23 active sentences using twochoice pictures. The subject and object of each picture are "stick-men" cartoons with round or square heads (e.g., "The circle paints the square"); (ii) 23 locative sentences using same geometrical shapes, circle and square (e.g., "The square is in the circle"). The sentences were presented in the written modality. KP performed these tasks competently; he scored 23/23 correct on the active sentences and 21/23 correct on the locative sentences. (c) Sentence-picture matching: written version [15]. The aim of this test is to assess the comprehension of written sentences involving different grammatical constructions. The test consists of 60 sentences and 60 sets of three pictures. The sentences are divided into four main types of constructions: reversible (e.g., "The dog is approaching the girl"); nonreversible (e.g., "The cat is licking the man"); gapped (e.g., "The girl is considering where to go") and converse relations (e.g., "The man is offering the money"). KP's performance was impaired for reversible sentences for which he obtained a score of 13 correct (controls" mean: 19.15). He performed creditably on two further categories (gapped sentences, KP: 12; controls: 15.3. Converse relations, KP: 5; controls: 6.85) and competently on the remaining class (non-reversible sentences, KP: 16; controls: 15.48). (d) Testjor reception of grammar ( TROG) [2]. This test assesses comprehension of grammatical contrast. The test stimuli consist of 80 sentences graded in grammatical complexity and 80 sets of four pictures. In the standard version of the test, the subject is asked to select the picture that corresponds to a spoken sentence'. KP was asked to select the picture that corresponded to a written
811
sentence. His performance was weak, he scored 63/80 correct (12/17 errors were made on locative sentences, e.g., "The pencil is on the box"). (e) The word-class comprehension test [6]. The aim of this test is to assess comprehension of grammatical word-classes within a written sentence context. The test consists of 105 fairly complex sentences in which, for the target word, there are three alternatives. Twenty-four sentences test prepositions (e.g., "The lady-in-waiting off/at/to the proud princess is dismissed by the queen"); 15 sentences test adjectives (e.g., "The serious/coloured/blunt lights at the Palladium were operated by the assistant producer"); 33 sentences test nouns (e.g., "The poultry farmers were plucking the feathers off the squawking mice/leopards/geese"). The remaining 33 sentences test verb tense (e.g., "The boy hits/was hit/were hit on the head by the falling stone"). KP had to select one of the three alternatives given in each sentence. Our patient's performance with the nouns was satisfactory (97% correct); his performance was adequate when he was asked to select adjectives (72% correct) but it was poor for verbs (57% correct) and at chance level for prepositions (33% correct). Overall, KP's comprehension of written sentences was fairly satisfactory with two exceptions. His comprehension of verb tense was impaired as was his comprehension of prepositions hut only in the context of complex sentences. A series of experiments were designed in order to investigate KP's verbal production and syntactic processing in more detail.
Experimentalinvestigation Test 1. Sentence production It was observed qualitatively that our patient had virtually no spontaneous speech. Our aim in this section therefore was to document more precisely KP's difficulties in producing sentences. Two tests were devised which assessed his ability to generate sentences. Sentence generation 1. The aim of this task was to investigate whether KP was able to generate sentences in a condition in which he could "use" a sentence structure given by the examiner to facilitate his own sentence production. Eighty sets of sentences together with pictures from the T R O G (see above) were presented to KP. The test was completed in three stages. First, the sentence structure was demonstrated by presenting a picture together with the appropriate sentence (e.g., "'The flower is longer than the pencil"). Second, KP was required to construct a sentence to describe a second picture using the same format given the prompt of the subject noun (e.g., "The k n i f e . . . " ) . Third, KP was required to con-
812
L. Manning and E. K. Warrington/Routes to naming
struct a sentence using the same format to describe a third picture without the prompt of the subject noun. Results. KP was unable to produce a single grammatically correct sentence, regardless of the presence or absence of a prompt. Whenever he attempted to generate a sentence his production was limited to naming parts of the picture (e.g., "The boy, the elephant is the b o y - push"; "the brown cat--the white cat"). Sentence generation 2. Our aim in this test was to investigate whether our patient was able to form sentences in a condition in which no word-retrieval was necessary, using a procedure similar to that reported by Von Stocker [36]. The test stimuli consist of 20 six- to nine-word written sentences from the T R O G . Each individual sentence was displayed in front of KP with the individual words in random order together with the corresponding picture. KP was required to re-arrange the words in order to obtain the sentence which described the picture. Three practice trials were given and no time limit was set. Results. KP was able to re-arrange 6/20 sentences correctly (e.g., correct sentence: "The boy is sitting but not eating"; e.g., an incorrect sentence: "The girl chases that is big the dog"). Comment. It has been demonstrated that KP's difficulties in producing sentences are not exclusively due to word-retrieval problems since his score did not improve significantly in the second condition in which wordretrieval was not required.
Test 2. Naming, reading and spelling objects and actions It was observed that our patient's action naming was markedly poor in comparison with his object naming. Our aim in this experiment therefore was to assess in more detail KP's ability to process verbs and nouns. The test stimuli consists of 75 objects/nouns (e.g., key, bridge, towel, zip) and 75 actions/verbs (e.g., read, leave, pat, skate) matched in frequency [30]. KP was required to name, read and spell each test item. The stimuli used in the naming test consisted of very clear and unambiguous line drawings selected from several published word-picture matching tasks. Naming and spelling were tested in an ABBA design and reading was tested in a subsequent session. Results. The percentage correct for verbs and nouns for each condition is presented in Table 1. The results
Table 1. KP's performance: Naming, reading and spelling the same pool of verbs and the same pool of nouns. Percentage correct
Naming Reading Spelling
n
Verbs %
Nouns %
75 75 75
19 99 83
75 99 83
show that his ability to name action/verb pictures was significantly worse than his ability to name object/noun pictures (Z2 = 54, P < 0.0001). His reading was set at ceiling for both nouns and verbs and there was a low but identical spelling error rate for nouns and verbs. Comment. It has been demonstrated that KP is able to read and spell nouns and verbs at a similarly competent level. However, there was a clear-cut word-class effect for naming.
Test 3. Action and object picture-word matching Our aim in this test was to investigate KP's comprehension of verbs and nouns using the word-picture matching test devised by McCarthy and Warrington [22]. The stimuli consist of 40 pairs of unambiguous line drawings, half of them are related pairs of objects (e.g., barbecue/casserole) and half are related pairs of actions (e.g., painting/drawing). The examiner spoke one of the items shown on each picture and KP had to point to the corresponding item. In order to probe both stimuli the picture-pairs were presented twice. Results. KP's performance was satisfactory for both, objects (97% correct) and actions (90% correct). Comment. The fact that KP's comprehension of semantically related verbs is preserved implies that his deficient action verb naming (experiment 2) was not due to degraded semantic knowledge of verbs.
Test 4. Concrete/abstract verb comprehension It was observed that KP was well able to perform the written version of a stringent test of verbal comprehension (Concrete/Abstract Comprehension Test; see above) including those few instances of verbs (e.g., hoist is it lift or carry?; sever is it cut or prune?; cogitate is it meditate or hesitate?). A concrete/abstract verb test was devised to confirm that his verb comprehension was equally intact. First, a pilot study was carried out to establish the concreteness rating of a pool of verbs. Six normal control subjects rated these verbs on a 7-point scale. This list of verbs was presented to four further controls in a twochoice format: a higher frequency synonym and a distractor item. On the basis of their results two lists of 25 verbs, one with a high concrete rating (e.g., confine does it mean imprison or end?) and one with a high abstract rating (e.g., acquiesce does it mean acquire or agree?) were selected. This written concrete and abstract 50 verb synonym test was presented to KP. Three controls matched in their reading IQ equivalent to KP were also tested. Results. The percentage correct for KP and the control
L. Manning and E. K. Warrington/Routes to naming subjects for the concrete and abstract verb comprehension are given in Table 2. The critical finding is that KP's performance on the abstract set was identical to that of the control subjects. Comment. KP's relatively preserved verb meaning, as documented in the concrete/abstract very synonym test, indicates that his impaired action naming cannot be attributed to impaired comprehension of verbs. This finding corroborates the results of the action word-picture matching test. Furthermore, his very satisfactory comprehension of abstract verbs indicates that the noun verb discrepancy cannot be accounted for in terms of a concrete/abstract dissociation.
Test 5. Written sentence-pieture matching test." verb tenses We have reported (see above) that KP had selective difficulties with verbs on a task where he was asked to choose the correct alternative within a sentence frame (the word-class comprehension test). We have also documented KP's relatively preserved ability to discriminate between semantically related verbs and to comprehend. The aim in this test is to investigate whether the disparity of results between impaired use of verbs and relatively preserved verb semantics could be due to a defective processing of verb tense. The test stimuli consist of 20 pairs of pictures, drawn from the Parisi and Pizzamiglio [31] set. Each pair shows unambiguously two tenses of the same action performed by the same actor (the actions were performed in the present, present simple, present progressive, past, past perfect and future tenses). A short sentence was written below the pairs of pictures (e.g., "The boy will draw"). KP was requested to match the written sentence to the appropriate picture. Results. KP's performance was at chance level: he scored 10/20 correct. For example, a picture of "the girl will drink" was matched with the picture of '~a girl drinking". ~'The boy has been running" was matched with "the boy is running". Comment. We have demonstrated that KP was no longer able to process verb tense; his performance was markedly impaired when he had to select a given tense from only two possibilities. This finding corroborates
Table 2. KP's and three controls' performance on the synonym concrete and abstract word comprehension test
Concrete Abstract
KP
Controls n -- 3 (mean)
19/25 20/25
22/25 20/25
813
KP's difficulties with verb tense as reported in the wordclass comprehension test.
Test 6. Sentence completion and con/rontation naming Breen and Warrington [4] have recently demonstrated that their anomic patient (NOR) who was insensitive to phonological and semantic cues improved his object naming significantly when a sentence frame was provided, i.e. he was able to provide the terminal noun which had been omitted in a sentence. Our aim in this experiment was to document KP's sentence completion skills. Clinical observation suggested that his abilities to complete a sentence with a high probability terminal word were very poor (see above). Two tests were devised: Sentence completion 1. Thirty sentences were constructed such that there was a medium or low probability that a particular word would be generated in the terminal position (e.g., ~'He went to the stationery shop to buy me s o m e . . 2' has a medium probability; "I was looking for t h e . . . " has a low probability). KP was required to read each sentence and generate a missing last word. Any plausible completion was accepted as correct. Results. The percentage correct for each level of probability was as follows: • Medium probability sentences: 46% correct; • Low probability sentences: 13% correct. Although KP's score was significantly better with the medium probability sentences than the low probability sentences 0~2 = 3.99, P < 0.05), his performance on this task was still quite impaired. Sentence completion 2. We have reported in test 2 (see above) KP's fairly adequate naming performance for objects (56/75 correct). In the present test, we re-presented the same 56 objects and the patient was required to name them. Additionally, for each object name there was a corresponding sentence frame with the target terminal noun omitted. In every case, the object name formed a high probability completion of the sentence (e.g., "The archer picked up his bow and fired a n . . . " ) . The sentences were written and no time limit was set. KP was asked to generate the terminal word. The two tasks, naming on visual confrontation and sentence completion, were presented in a ABBA design. Results. The few plausible responses which did not correspond to our target word were accepted as correct. KP's performance on the sentence completion task was poorer (34% correct) than his performance on visual confrontation naming (89% correct). There was a highly significant difference in retrieving the object's name in the two conditions (;~2 = 47, P < 0.0001). Comment. We have demonstrated in the first task that KP had great difficulty in completing low probability sentences and his performance showed only moderate improvement with medium probability sentences. More importantly, we have demonstrated in the second task a clear dissociation between noun-retrieval by visual confrontation which appears relatively well preserved and
814
L. Manning and E. K. Warrington/Routes to naming
noun-retrieval through sentence completion which is markedly impaired.
Discussion
We have presented a patient in whom the following findings were documented: KP had markedly defective sentence generation. He showed a clear-cut grammatical class dissociation with action naming significantly worse than object naming. Although he was able to name objects on visual confrontation, he was unable to retrieve those very same object names on a sentence completion task in which he was provided with a high probability frame to elicit the terminal noun. It was also demonstrated that he was able to comprehend the meaning of verbs on a word-picture matching task and on a more stringent synonym concrete/abstract verb comprehension test. However, his preserved knowledge of the semantics of verbs contrasted with his inability to process the syntax of verbs, as it was demonstrated by his very impaired performance on verb tense tasks. KP's sentence comprehension on the written modality was generally adequate, except for the following grammatical constructions: his performance on prepositional verb sentences was weak in complex constructions. He was able to comprehend reversible sentences in constructions in which animacy/saliency was neutralised. Sentences in which verb tense was crucial were poorly understood. First, we discuss our patient's deficient sentence generation. Second, we examine his verb processing in relation to both his impaired sentence generation and his inability to use high probability sentence frames to produce nouns which he was able to produce on visual confrontation. The patient's inability to generate sentences occurred within the context of a relatively well-preserved wordretrieval ability on visual confrontation tests. KP's failure to generate sentences does not seem to be due to the characteristics generally attributed to "dynamic aphasia" either in Luria's [19, 21] or Costello and Warrington's [5] account of this syndrome. The former authors stated that dynamic aphasia is due to a marked disturbance in spontaneous expanded speech, all other aspects of language remaining intact. Costello and Warrington argued that it is the incapacity to organise or plan thoughts which causes the symptoms of dynamic aphasia, rather than the inability to implement that planning. KP did indeed seem unable to organise or plan thoughts. However, his inability to process verb tenses together with his difficulties in complex locative prepositional sentences and some reversible sentences discourage any attempt to account for his sentence generation impairment in terms of dynamic aphasia. By the same token, his impaired syntactic processing leads us to consider some evidence on grammatical issues. Grammatical difficulties in sentence production were classified in two different types of agrammatism, m o r -
phological and syntactic, by Tissot et al. [35]. The morphological agrammatic patients would have problems with function words and word endings. KP presented with both classes or errors. The syntactic agrammatic patients would have difficulties with the word order and the retrieval of verbs. KP had difficulties with word order when he was asked to organise written words into sentences. However, his speech was so sparse that it was hardly possible to detect spontaneous word order difficulties. He also had difficulties in choosing verb tenses. There are two main accounts of sentence generation, Garrett's model [9, 10] and Bock's model [3]. We will comment briefly on these two models and consider which, if either, captures KP's deficit in sentence production. Garrett's model postulates a division between lexical retrieval and the generation of a functional argument structure, both of which belong in the "message-level representation". In cases in which the lexical retrieval ability is preserved but the possibility of creating a functional predicate/argument structure is impaired, the patients would be left with a collection of content words but no predicate/argument structure in which to insert them. However, KP's inability to generate sentences cannot be explained by this model. Although he was able to name single objects, he had a grammatical class dissociation. KP's impaired action naming does not seem to be readily explained by a theoretically preserved lexical retrieval function which makes no grammatical-class specifications. Bock [3] proposed that verb retrieval is crucial in sentence generation because verbs carry the information necessary to assign thematic roles. A verb's specification of functional relations is responsible for the order in which the semantically specified and grammatically tagged words will be linked together. In Bock's sentence generation model, input from several levels may contribute to the process of generating sentences, and at each stage verbs play a central role in subsequent aspects of sentence production. KP's defective verb tense processing and action naming, therefore, must be implicated in his inability to generate sentences. Although the terms noun and verb refer to syntactic classes, they elicit meaning differences. Evidence has emerged that there are psychological distinctions in the processing and representation of nominal and verb meaning [33]. Besides the well-known dissimilarities between verbs and nouns in terms of the former being less memorable, harder to interpret and possessing a wider range of meanings across languages, the functional differences are more relevant in our attempt to interpret our patient's performances. KP's pattern of performance in relation to verbs suggested a dissociation between, on the one hand, the use of rules that command the syntactic aspects of verbs, more specifically, verb tenses and, on the other hand, the comprehension of verb meaning including some fairly infrequent instances of verbs. This finding, with regard to his impaired sentence generation would imply that
L. Manning and E. K. Warrington/Routes to naming comprehension of verbs (relatively preserved in KP) is not sufficient to generate sentences and that adequate verb conjugation is crucially necessary. This suggestion is parallel to that advanced by several authors who studied other syntactic constructions. For example, Nespoulous et al. [28] studied a patient whose difficulties were exclusively observed in the production of sentences due to deficits at the level of "use" of morphology rather than defective syntactic knowledge. A further subtle dissociation was studied by Linebarger et al. [18]. They came to the conclusion that there may be a dissociation between the ability to perceive syntax (the patients detected over 90% of grammatically wrong constructions) and the ability to comprehend syntax (they scored at chance level on picture-pointing tests in which word order was crucial). It is perhaps not surprising then, that patients whose semantic and syntactic verb processing is spared have a preserved performance on sentence generation tasks despite anomic difficulties. What appears to be more interesting is the finding that verb syntax is necessary in order to make use of sentence frames in naming tasks. Noun production in sentence completion tasks seems to be the result of verbal activations that are not present in other forms of verbal prompting such as, for example, definitions nor in naming from visual confrontation. Zingeser and Berndt [45] described a patient (HY) with marked naming difficulties whose performance on sentence completion tasks was virtually error free (95% correct on high probability sentence frames versus 70% correct on naming on visual confrontation). HY's verb production was far superior to noun production in single word naming (62.5% naming action pictures in contrast to 30-35% naming objects). A further case report with very similar findings is the study by Breen and Warrington [4]. Their patient NOR's naming skills were markedly impaired. Phonological and semantic cues were ineffective in helping him retrieve a given noun. In contrast, a sentence frame, independent of the level of probability to elicit a target had a significant facilitatory effect. Thus, the patient named 76% of pictures when presented together with an incomplete sentence versus 29% correct on visual confrontation tasks without sentence frames. NOR's grammatical-class dissociation was similarly clear-cut: he could name only 15% of the coloured pictures of objects but 68% of pictured actions. Zingeser and Berndt's and Breen and Warrington's data suggest an integral association between preserved action naming and the ability to use sentential frames in order to retrieve nouns which are not retrieved in a confrontational naming task. KP presented with the converse pattern of performance: deficient action naming and inability to use sentence frames to retrieve nouns which he was able to retrieve in an object naming task. We would like to interpret this case following the argument put forward by Breen and Warrington [4] who argued for there being two routes to phonology, the nominal and the propositional routes.
815
It is interesting to recall that as early as 1876, Hughlings Jackson [13] distinguished between propositional speech and automatic speech. This distinction has been elaborated and refined by Head [11], Pick [32] and Kleist [16] who in 1934 used the already well-established distinction between "nominalising" speech and "propositionising" speech. In linguistics, the term ~'nominalising" corresponds to the paradigmatic speech which is used in naming and the term "propositionising" corresponds to the syntagmatic speech which refers to the act of producing sentences in which words are modified by the context [14]. Similarly, Luria [19] differentiated between nominative and propositional language. Breen and Warrington emphasised the possibility that noun-retrieval is the result of two different procedures: noun-retrieval in propositional speech and in nominal contexts. They interpreted their case, NOR, who was very impaired in confrontation naming tasks but obtained a significant benefit from sentence contexts, in terms of the operation of an on-line, integrative language processor. Characteristically, this processor is thought to involve a shortterm memory system and to operate both at a level of processing complexity beyond single words, and in an anticipatory mode. In other words, the language processor constructs propositional speech [23, 25], This process would constitute one of the two routes to phonology, the other route being one which involves the procedures of naming to confrontation. Following their hypothesis, KP's relatively preserved performance on naming tasks would indicate a preserved nominal route whereas his marked inability to generate sentences and, in particular, his incapacity to benefit from sentence frames to name an object would be consistent with such a damaged integrative language processor, the other route to phonology. Finally, it has been documented that: (i) verbs appear to be crucial in sentence generation; (ii) the provision of sentence frames had highly beneficial facilitatory effects in the production of a content word in patients whose action naming was better preserved than their object naming; (iii) our patient presented with impaired action naming and defective verb syntax in the context of relatively preserved verb semantics: and (iv) KP was unable to benefit from the provision of sentence frames in producing a noun. From the perspective of the notion of two routes to naming, these data suggest that verb syntax and/or action naming are crucial for "'using" sentences as verbal prompters, i.e. the propositional route probably hinges on intact or nearly-perfect verb processing. The obvious next stage is to study the effects on naming via propositional speech of selectively impaired semantics of verbs, syntax of verbs and action naming. In summary, our patient presented with a pattern of performance which provides sufficient information to be interpreted in terms of a double dissociation vis-Zt-vis both Zingeser and Berndt's, and Breen and Warrington's patients: KP's defective action naming, on the one hand, contrasted with both HY's and NOR's verb production which was far superior to noun production in single word
816
L. Manning and E. K. Warrington/Routes to naming
naming. Moreover, KP's marked inability to benefit from sentence frames to retrieve a highly predictable terminal noun was the exact opposite to HY's and NOR's performance.
Acknowledgement--We are most grateful to Dr A. J. Lees for permission to study a patient under his care and to report our findings.
References
1. Berndt, R. Symptoms co-occurrence and dissociation in the interpretation of agrammatism, In The Cognitive Neuropsychology of Language, M. Coltheart, G. Sartori and R. Job (Editors), pp. 221 233. Lawrence Erlbaum Associates, London, 1987. 2. Bishop, D. Test for the Reception of Grammar. Medical Research Council. Chapel Press, England, 1990. 3. Bock, K. J. Co-ordinating words and syntax in speech plans. In Progress in the Psychology of Language, A. Ellis (Editor). Lawrence Erlbaum Associates, London, 1987. 4. Breen, K. and Warrington, E. K. A study ofanomia: Evidence for a distinction between nominal and propositional language. Cortex 30, 231-245, 1994. 5. Costello, A. de L. and Warrington, E. K. Dynamic aphasia. The selective impairment of verbal planning. Cortex 25, 103-114, 1989. 6. Coughlan, A. The Word Class Comprehension Test. Unpublished. 7. De Bleser, R. From agrammatism to paragrammatism: German aphasiological traditions and grammatical disturbances. Cognitive Neuropsychology 4, 187-256, 1987. 8. Freud, S. On Aphasia: A Critical Study. (Translated by E. Stengel.) International Universities Press, New York, 1891. 9. Garrett, M. F. Levels of processing in sentence production. In Language Production. Vol. 1. Speech and Talk, B. Butterworth (Editor). Academic Press, London, 1980. 10. Garrett, M. F. Production of speech: Observation from normal and pathological language use. In Normality and Pathology in Cognitive Functions, A. Ellis (Editor). Academic Press, London, 1982. 11. Head, H. Aphasia and Kindred Disorders of Speech. Cambridge University Press, Cambridge, 1926. 12. Heilbronner, K. Uber Agrammatismus und die Storung der inneren Spraghe. Archivfiir Psychiatrie und Nervenkrankheiten 41,653 683. 13. Hughlings Jackson, J. 1876. Reprinted in Selected Writings of John Hughlings Jackson, J. Taylor (Editor). Hodder and Stoughton, London, 1932. 14. Jacobson, R. Studies on child language and aphasia. Janua Linguarum Series Minor 144, Mouton, The Hague, 1971. 15. Kay, J., Lesser, R. and Coltheart, M. PALPA. Lawrence Erlbaum Associates, London, 1992. 16. Kleist, K. Gehirnapathologies. Barth, Leipzig, 1934. 17. Lichtheim, L. On aphasia. Brain 7, 433-484, 1885.
18. Linebarger, M. C., Schwartz, M. F. and Saffran, E. M. Sensitivity to grammatical structure in so-called agrammatic aphasics. Cognition 13, 361-392, 1983. 19. Luria, A. Traumatic Aphasia. Mouton, The Hague, 1970. 20. Luria, A. and Hutton, J. T. A modern assessment of the basic forms of aphasia. Brain and Language 4, 129-151, 1977. 21. Luria, A. and Tsvetkova, L. S. The mechanisms of dynamic aphasia. Foundations of Language 4, 296307, 1968. 22. McCarthy, R. A. and Warrington, E. K. Category specificity in an agrammatic patient: The relative impairment of verb retrieval and comprehension. Neuropsychologia 23, 709-727, 1985. 23. McCarthy, R. A. and Warrington, E. K. Understanding: A function of short term memory? Brain 110, 1565-1578, 1987. 24. Manning, L. and Campbell, R. Optic aphasia with spared action naming. A description and possible loci of impairment. Neuropsychologia 6, 587-593, 1992. 25. Marslen-Wilson, W. D. and Tyler, L. K. The temporal structure of spoken language comprehension. Cognition 6, 1-71, 1980. 26. Miceli, G., Silveri, M. C., Villa, G. and Caramazza, A. On the basis for the agrammatic's difficulty in producing main verbs. Cortex 20, 207-220, 1984. 27. Nelson, H. E. and Willison, J. R. The National Adult Reading Test. The NFER-Nelson Publishing Co., Windsor, England, 1991. 28. Nespoulous, J. L., Dordain, M., Peron, C., Ska, B., Bub, D., Caplan, D., Mehler, J. and Roch Lecours, A. Agrammatism in sentence production without comprehension deficits: Reduced availability of syntactic structures and/or of grammatical morphemes? A case study. Brain and Language 33, 273 295, 1988. 29. Oldfield, R. C. and Wingfield, A. Response latencies in naming objects. Quarterly Journal of Experimental Psychology 17, 273 281, 1965. 30. Orpwood, L. and Warrington, E. K. Word specific impairments in naming and spelling but not reading. Cortex 31,239-265, 1995. 31. Parisi, D. and Pizzamiglio, L. Syntactic comprehension in aphasia. Cortex 6, 204-215, 1970. 32. Pick, A. 1931. Aphasia. Reprinted in translation by Brown J. W. Charles C. Thomas, Springfields, IL, 1973. 33. Reyna, V. F. Understanding verbs: Easy extension, hard comprehension. In Progress in the Psychology of Language, A. Ellis (Editor). Lawrence Erlbaum Associates, London, 1987. 34. Schwartz, M. F., Saffran, E. M. and Marin, O. S. M. The word order problem in agrammatism. I. Comprehension. Brain and Language 10, 249 262, 1980. 35. Tissot, R., Mounin, G. and Lhermitte, F. L'Agrammatism. Dessart, Paris, 1973. 36. Von Stockert, T. R. Recognition of syntactic structure in aphasic patients. Cortex 8, 323-354, 1972. 37. Warrington, E. K. Recognition Memory Test. The NFER-Nelson Publishing Co., Windsor, England, 1984. 38. Warrington, E. K. and Clegg, F. Recognition Memory Test, Shortened Versions. Unpublished.
L. Manning and E. K. Warrington/Routes to naming 39. Warrington, E. K., Jackson, M., Orpwood, L. and McCarthy, R. A. Concrete and Abstract Word Comprehension Test. Unpublished. 40. Warrington, E. K. and McCarthy, R. A. Category access dysphasia. Brain 106, 859-878, 1983. 41. Warrington, E K. and Shallice, T. Category specific semantic impairments. Brain 107, 829 853, 1984. 42. Wernicke, C. Nervenheilkunde. Die neueren Arbeinten uber Aphasie. Fortschritte der Medizin 824~830, 1885.
817
43. Williams, S. and Canter, G. The influence of situational context on naming performance in aphasic syndromes. Brain and Language 6, 63-74, 1982. 44. Williams, S. and Canter, G. Action naming performance in four syndromes of aphasia. Brain and Language 32, 124-136, 1987. 45. Zingeser, L. and Berndt, R. Grammatical class and content effects in a case of pure anomia: Implications for models of lexical processing. Cognitive Neuropsycholo,qy 5, 473-516, 1988.