Brain, Vol. 124, No. 8, 1533-1543,
August 2001
© 2001 Oxford University Press
Refractory dyslexia
Evidence of multiple task-specific phonological output stores
Dementia Research Group, Department of Clinical Neurology, Institute of Neurology, London, UK
Correspondence to:
Professor Elizabeth K. Warrington, Dementia Research Group, The National Hospital for Neurology and Neurosurgery, 8–11 Queen Square, London WC1N 3BG, UK E-mail: e.warrington{at}dementia.ion.ucl.ac.uk
 |
Abstract |
|---|
 Top Abstract IntroductionCase reportNeuropsychological backgroundDocumentation of reading skillsExperimental investigationsDiscussionAcknowledgementsReferences |
|---|
We investigated the case of a patient whose reading was characterized by multiple phonemic paraphasic errors. An error analysis of a large corpus of reading responses (758 words, 86 non-words) highlighted the preponderance of phonological errors which did not occur in his naming, repetition or spontaneous speech. His comprehension of the written word was relatively preserved, even for words he was unable to read aloud. We suggest that his impairment lies at the level of the phonological output store. We also demonstrate that his reading performance was facilitated by increasing the response–stimulus delay. The strong influence of temporal factors is shown to be task-specific. Two main points are drawn from our results. First, we argue that our patient can be characterized as having a refractory access type of deficit; to our knowledge, no previous case of a refractory deficit affecting word reading has been reported. Secondly, the task specificity of both the phonological error pattern and the sensitivity to temporal factors is difficult to reconcile with the idea of a unitary phonological output store. Contrary to orthodox neuropsychological models, we propose that there are independent stores specific for reading and spoken output.
access disorders; refractoriness; phonology; output stores; task-specific
BPVS = British Picture Vocabulary Scale; NART = National Adult Reading Test; PIQ = performance IQ; RSI = response–stimulus interval; VIQ = verbal IQ; WAIS-R = revised Wechsler Adult Intelligence Scale
|
Introduction |
|---|
|
TopAbstractIntroductionCase reportNeuropsychological backgroundDocumentation of reading skillsExperimental investigationsDiscussionAcknowledgementsReferences |
|---|
Cognitive neuropsychological models of language propose separate routes for processing the spoken and written word (Morton and Patterson, 1980
). Furthermore, recent models describe distinct input and output processes for each of these cognitive skills, with modality-specific input (visual and auditory) and output (phonological and orthographic) lexicons (Morton, 1979, 1980). However, the phonological and orthographic output stores retain an organized structure. Word variables such as word frequency and grammatical class are seemingly represented at the level of the post-semantic output stores for both reading (Coltheart, 1980; Caramazza and Hillis, 1990, 1991) and spelling (Baxter and Warrington, 1985; Caramazza and Hillis, 1991). The organization of the output stores has been examined further with data from a patient with preserved reading but impaired naming and spelling (M.R.F. in Orpwood and Warrington, 1995). This dissociation between reading and naming was held to affect the output stores in as far as word comprehension was intact. Contrary to the orthodox view, this selective preservation of lexical reading despite apparent damage to the phonological output store was interpreted as evidence of two output phonologies. This view has been challenged by those who suggest that this pattern of deficits merely reflects inherent differences between the naming and reading processes (Lambon-Ralph et al., 1999). They claim that the naming process can be facilitated by other forms of phonological information and that a common phonological store subserves multiple tasks including naming and reading. Indeed, they predict that `naming and reading will create a single, not a double dissociation' (Lambon-Ralph et al., 1999, p. 167). Impaired performance on word production tasks may indicate that the central representation itself has been damaged or that access to an intact representation has been compromised. Impaired access to phonological or orthographic representations causes the information stored within to become temporarily unavailable. One consequence of such restricted access is response inconsistency. Access disorders were first described in a patient with acquired dyslexia on the basis that from trial to trial his response consistency on single word reading tasks was very low (A.R. in Warrington and Shallice, 1979). A.R. also showed evidence of priming effects in reading and, to a lesser extent, naming. These findings were considered to be incompatible with damage to the central representation but are predicted by the access disorder account.
In this first account of an access disorder, temporal factors were not considered. The concept of an `access' disorder, in the sense that access can be unreliable or inconsistent, was developed and extended in the context of verbal comprehension deficits. The role of temporal factors came to light with evidence that the introduction of even a short delay after each response may facilitate performance (Warrington and McCarthy, 1983, 1987; Cipolotti and Warrington, 1995; McNeil et al., 1994). In these cases though, the facilitatory effect of increasing delay between response and presentation of the next stimulus (response–stimulus delay) was limited to retrieval of information from specific semantic categories. To account for the temporary unavailability of semantic information in these cases, the term refractoriness was introduced. Defined as `the reduction in the ability to utilise the system for a period of time following activation' (Warrington and McCarthy, 1983, p. 874), `refractoriness' describes a state in which access to semantic information is not only inconsistent but also dependent upon temporal factors. It should be noted, however, that not all cases with temporary inconsistency or a comprehension deficit show sensitivity to temporal factors (Warrington and Cipolotti, 1996; Warrington and Leff, 2000).
Although the majority of cases showing refractoriness have done so in the context of comprehension deficits, two cases of a refractory anomia have been reported (Wilshire, 1996, unpublished Ph.D. thesis, Cambridge; F.A.S. in McCarthy and Kartsounis, 2000). In the first case, the patient showed relatively preserved comprehension, but naming performance was affected by temporal factors. In the second case, the patient was particularly remarkable for demonstrating a refractory disorder of word retrieval which spared word comprehension. Unlike the majority of access disorder patients, F.A.S.'s word retrieval refractoriness cannot be attributed to a deficit in semantic access. This raises the possibility that the dynamics of any module of cognitive processing can become refractory as well as unavailable.
Here we describe a patient whose reading is characterized by multiple phonological errors. We report a series of experiments which demonstrate that this phonological impairment is task-specific. In particular, spontaneous propositional speech, though characterized by grave word-finding difficulties, is not subject to phonological disintegration. A further series of experiments shows that the effect of varying the temporal factors is specific to reading alone. It is argued that these data are difficult to reconcile with the idea of a unitary phonological output store. These phenomena also provide previously undocumented evidence of temporal factors affecting the reading process.
|
Case report |
|---|
|
TopAbstractIntroductionCase reportNeuropsychological backgroundDocumentation of reading skillsExperimental investigationsDiscussionAcknowledgementsReferences |
|---|
V.Y.G. was a 64-year-old ambidextrous retired mechanic and farmer, admitted to the National Hospital for Neurology and Neurosurgery for investigation of language and memory problems. MRI revealed localized atrophy of the left temporal lobe, diffuse temporal stem lacunes which were considered insufficient to explain the temporal lobe atrophy, and widespread vascular lesions (see Figs 1 and 2
). The patient fulfilled the criteria for progressive non-fluent aphasia (Neary et al., 1998) with co-existent vascular disease.
|
|
|
Neuropsychological background |
|---|
|
TopAbstractIntroductionCase reportNeuropsychological backgroundDocumentation of reading skillsExperimental investigationsDiscussionAcknowledgementsReferences |
|---|
On formal examination, V.Y.G. was assessed on a short version of the revised Wechsler Adult Intelligence Scale (WAIS-R; Warrington et al., 1986) and obtained a verbal IQ (VIQ) of 65 and a performance IQ (PIQ) of 107. V.Y.G. showed selective difficulties with verbal recognition memory on the short form of the Recognition Memory Test for words (Warrington, 1984
). V.Y.G. was able to solve simple tests of frontal lobe executive function (alternating bimanual movements; Weigl Sorting Test; Weigl, 1927), but experienced difficulties with more demanding tasks (Brixton; Burgess and Shallice, 1996). V.Y.G. demonstrated a marked acalculia (GDA 0 out of 24; Jackson and Warrington, 1986) and agraphia (GDS 0 out of 24; Baxter and Warrington, 1994) and was only able to write his own name.
V.Y.G.'s propositional speech, though characterized by considerable word-finding difficulties, was fluent and voluble and produced with appropriate prosody. Indeed it was noted that in two instances V.Y.G. was able to mimic an accent in attempting to name a picture; an American accent in the case of a stetson (cowboy) `hat', and an upper class accent when circumlocuting around the target `bow tie'. His sentence structures, however, were marred by frequent circumlocutory phrases to compensate for his very obvious word-finding difficulties. However, there was no suggestion of phonemic or neologistic errors. His word retrieval abilities were severely impaired on tests of naming common objects [Oldfield Naming Test four out of 30, Graded Naming Test 0 out of 30 (McKenna and Warrington, 1983)]. Naming performance on the Oldfield objects was not facilitated by introducing a delay between the presentation of items (McCarthy and Kartsounis, 2000). With a 5 s delay between response and presentation of the next stimulus, his score remained four out of 30. His ability to repeat spoken words was relatively preserved, but showed an interaction between word length and frequency. On three sets of 30 high frequency words of increasing length, V.Y.G. correctly repeated 29, 28 and 22 words, respectively, whereas with three length-matched sets of low frequency words repetition performance decreased to 26, 22 and 18 correct responses. Repetition of clichés and sentences was severely impaired. Comprehension of the low frequency words was found to be somewhat impaired. V.Y.G. scored 110 out of 150 on the long form of the British Picture Vocabulary Scale (BPVS) and 17 out of 25 on a test of Concrete Synonyms, a score close to the 10th percentile. Auditory word perception was within normal limits (35 out of 40; Coughlan and Warrington, 1978).
V.Y.G.'s reading skills were severely impaired, managing to read correctly only one out of 50 items on the National Adult Reading Test (NART; Nelson, 1982). The majority of errors on the NART were of a phonemic or neologistic nature, though six clear regularizations were also observed. V.Y.G. also made many phonemic, neologistic and regularization errors in reading a short prose passage, taking 2 min 11 s to do so. On the Glushko corpus of 86 regular and exception non-words (Glushko, 1979), he achieved a score of 59 out of 86. Of these, 16 were lexicalizations.
In summary, V.Y.G.'s language capacities appear to bear the brunt of his cognitive impairment, with his spontaneous speech being fluent yet circumlocutory. His dyslexia, characterized by numerous phonemic errors, is especially notable, particularly given the relative absence of phonemic errors in propositional speech. Our investigation focuses on documenting his dyslexic syndrome in greater detail and comparing his performance on comparable tests of reading, repetition and word retrieval.
|
Documentation of reading skills |
|---|
|
TopAbstractIntroductionCase reportNeuropsychological backgroundDocumentation of reading skillsExperimental investigationsDiscussionAcknowledgementsReferences |
|---|
Test 1: regular and irregular word reading
In reading the NART and the short prose passage during the initial assessment, V.Y.G. made many phonemic errors (e.g. psalm
sparm, nauseaorsind), but also several clear regularization errors (e.g. chord, thyme). Although V.Y.G.'s reading is clearly impaired, most of the items on the NART are low frequency. In order to focus on the nature of his dyslexia, we wished to examine whether word frequency or word regularity factors affected the type or frequency of his errors.
The stimuli consisted of the Patterson and Hodges corpus of 192 monosyllabic words, which combine regularity/irregularity and high/low frequency in a factorial design (Patterson and Hodges, 1992). The words were presented in list form with a column of 15 words per page. The investigator indicated each word to the patient in turn in order to ensure he was reading the correct word.
The number of reading errors in each frequency and regularity category is given in Table 1. There is a significant effect of both frequency and regularity (
2 = 6.9 and 2 = 6.9, P < 0.01 for both).
|
His reading responses were classified as follows:
- (i) Initial phonological errors. The substitution, omission or addition of the initial phoneme in the target word (e.g. mugpug, coilgoil) irrespective of whether the initial error was followed by an additional error on the middle or final phoneme (e.g. globeclode).
- (ii) Middle/final phonological errors. The substitution, omission or addition of the middle or final phonemes in the target word (e.g. starch
starge, bulkbulp).
- (iii) Regularization errors. The substitution of a common grapheme–phoneme correspondence in an irregular word.
- (iv) Miscellaneous. Errors with less than two phonemes in common with the target (e.g. youth
dus, soakdouch).
- (ii) Middle/final phonological errors. The substitution, omission or addition of the middle or final phonemes in the target word (e.g. starch
The number of each error type and the percentage error rate are given in Table 2. In only six instances in this error corpus could the error response be considered as a perseveration of the previous target [e.g. whileswhy (previous target: sure)].
|
Test 2: comprehension of the written and spoken word
Test 1 above demonstrates the great difficulty experienced by V.Y.G. in reading single words. In assessing the locus of his deficit, it is essential to establish whether V.Y.G. can comprehend the written and spoken word.
The short form of the BPVS was administered in both written and oral form on two separate occasions. For the written form, target words were presented on separate word cards. V.Y.G. was not required to read the word aloud, but merely to point to the appropriate picture.
V.Y.G. obtained a score of 27 out of 32 on the written form and 29 out of 32 on the oral form of the test. His performance on the written version of this task seemed to be reasonably competent.
Test 3: comprehension of written irregular words
On the written form of the BPVS, V.Y.G. demonstrated that his comprehension of the written word remained relatively intact. Our aim in this test was to examine explicitly whether V.Y.G. is able to comprehend words which he cannot read aloud. In order to minimize access to semantic information via the phonological route, we tested V.Y.G.'s ability to read and comprehend irregular words.
The test stimuli were the Funnell set of 25 written irregular target words presented on separate cards (Funnell, 1983). Comprehension was tested using a three choice word–picture matching task (target, semantic distracter and unrelated distracter). V.Y.G. was requested to read each word aloud and then to point to the appropriate picture. Immediately following completion of the set, he was required to read aloud the words again for a second and third trial.
On the word–picture matching task V.Y.G. scored 24 out of 25, whilst on the word reading task he obtained scores of 18 out of 25, nine out of 25 and nine out of 25 on the three trials, respectively. This provides clear evidence that V.Y.G. can comprehend words which he misreads aloud. A consistency analysis was also calculated from his reading responses, based on his overall error rate. His error pattern was found to be inconsistent, conforming closely to the expected values of the binomial distribution [2(1) = 0.1, P > 0.5].
|
Experimental investigations |
|---|
|
TopAbstractIntroductionCase reportNeuropsychological backgroundDocumentation of reading skillsExperimental investigationsDiscussionAcknowledgementsReferences |
|---|
Experiment 1: reading versus naming versus repetition
From the neuropsychological assessment, it is apparent that V.Y.G. has a mild repetition deficit and a grave anomia. In this experiment, we directly compared V.Y.G.'s reading, naming and repetition in order to consider whether his impoverished naming and repetition skills show the same pattern of phonological errors which characterizes his reading.
Procedure
The test stimuli were 40 nouns and 40 verbs matched for frequency (Thorndike-Lorge). Reading and naming were tested in blocks of 40 words (20 nouns and 20 verbs) using an ABBA design. Later in the testing session, V.Y.G. was requested to repeat the 80 target items. The presentation rate was at the natural pace of the patient.
Results
The number of errors and percentage error rates for each presentation condition are given in Table 3. Reading performance was significantly less impaired than naming (2 = 8.44, P < 0.01) but also significantly worse than repetition performance, which was almost at ceiling (2 = 32.41, P < 0.001).
|
The reading and naming errors were classified as phonological, circumlocutory, semantic or miscellaneous. The number of each type of error for each task is given in Table 4
. Miscellaneous errors included omissions, gestures and generalizations. In contrast to the multiple phonological errors in his reading, the majority of V.Y.G.'s naming errors were circumlocutions. Apart from a small number of semantic and phonological errors, other errors in naming consisted mainly of omissions, gestures or generalizations.
|
Comment
This experiment demonstrates that the phonological errors observed in V.Y.G.'s reading are not present in his naming or repetition. This suggests that the pattern of phonological errors which characterizes V.Y.G.'s reading is task-specific.
Experiment 2: effect of temporal factors upon reading
It was observed on the initial assessment of V.Y.G.'s ability to read that his performance was affected by the rate at which he produced responses. The number of errors on word list reading tasks appeared to be reduced by pacing V.Y.G. rather than allowing him to read at his own faster speed. In this experiment, we documented this apparent effect of temporal factors by introducing a fixed delay after each reading response.
Procedure
The test stimuli were eight blocks of 15 high-frequency Thorndike–Lorge words. The sets were matched for word length, each containing five three-letter, five five-letter and five seven-letter words. There were two temporal conditions: a response–stimulus interval (RSI) of 2 and of 5 s. The stimuli were presented on two occasions. The RSI was alternated in an ABAB design on the first occasion and in a BABA design on the second occasion.
Results
The percentage error rates for reading at each RSI are shown in Table 5. Reading performance is greater at the 5 s than at the 2 s RSI in every block except block five. A Sign test on the differences shows this temporal condition effect to be significant (z = 4.17, P < 0.001). There is also a significant word length effect which is greater at the 2 s than at the 5 s RSI (2 = 24.76, P < 0.001 and 2 = 9.81, P < 0.01, respectively).
|
An error analysis of performance at each RSI is given in Table 6
. Although the total error rate was higher with the short delay, the distribution of error types in each temporal condition was approximately equivalent. Again phonological errors predominated.
|
Comment
In this experiment, we have demonstrated that V.Y.G.'s reading is strongly influenced by the RSI. This suggests that under optimal conditions, his ability to read single words may be better than is apparent under normal clinical testing conditions.
Experiment 3: effect of temporal factors upon reading and repetition
In the previous experiment, the RSI was shown to have a significant effect on the accuracy of V.Y.G.'s reading. Here we consider whether temporal factors affect repetition performance as well as reading.
Procedure
In an effort to prevent a ceiling effect in V.Y.G.'s repetition performance (see Experiment 1), the words used in this experiment were not the short high frequency words used in Experiment 2. Instead, the test stimuli were 30 high frequency and 30 low frequency tri-syllabic words (McCarthy and Warrington, 1984). There were two RSIs: a slow RSI of 5 s and a fast RSI of 2 s. The stimuli were presented in two phases. First, V.Y.G. was requested to read or repeat the words in an AB design, with the RSI varied in a corresponding ABBA design. Subsequently, the stimuli were presented in the reverse pattern with V.Y.G. reading or repeating in a BA design whilst the RSI was again varied in the ABBA fashion.
Results
The percentage error rates for reading and repetition at different RSIs are shown in Table 7. A Sign test of the differences shows a significant effect of temporal factors on reading, confirming the results of Experiment 2 (x2 = 4, n = 12, P < 0.05, one-tailed). However, no such significant effect was found for repetition (x2 = 6, n = 15, P > 0.3, one-tailed). With the longer tri-syllabic words, repetition performance falls below ceiling but, consequently, reading performance is also poorer than in previous experiments.
|
Comment
The apparent floor effect in reading performance may cause the temporal effect to be less strong in this experiment than in the last. Nevertheless, this experiment has confirmed the finding of Experiment 2 that temporal factors do have an effect upon V.Y.G.'s reading, and has shown that temporal factors do not have a significant influence upon the accuracy of V.Y.G.'s repetition. As noted earlier, naming performance on the Oldfield Naming test was not facilitated by introducing a delay after each response. Thus it appears that the effect of manipulating the RSI is indeed restricted to the modality of reading.
Experiment 4: comparison of reading and talking
From our clinical assessment, it was apparent that V.Y.G.'s spontaneous propositional speech was to a large extent free from the phonological errors which pervade his reading. In order to quantify the frequency with which errors were made in spontaneous speech, a sample corpus of 1000 words extracted from spontaneous unstructured conversation was analysed for word errors. Fourteen phonemic errors were noted, equivalent to an error rate of 1.4%. This error rate is strikingly lower than for reading nouns and verbs (49%, Experiment 1) or regular and irregular words (65%, Test 1).
In our preliminary analysis, we observed that V.Y.G. appears to produce far fewer phonemic errors in spontaneous speech than reading. However, the samples of words analysed were not matched for frequency, regularity or word length. A higher error rate on reading tasks than in spontaneous speech would be predicted if targets in the reading task fell outside of an impoverished vocabulary, to which spontaneous speech was restricted. We examined this possibility with a direct matched comparison of speech and reading.
Procedure
Four short samples of relatively meaningful spontaneous conversational speech were transcribed, which V.Y.G. subsequently was requested to read aloud. The time taken to read each set of sentences was also recorded and compared with the original duration of the utterance.
Results
The number of errors, the time taken and the relative word production rate for each set of sentences are given in Table 8. The results demonstrate that V.Y.G. makes phonological errors reading words which are correct in his spoken vocabulary. Of the 57 words read incorrectly or omitted, the majority of errors were phonemic substitutions or deletions and two were semantic substitutions (e.g. sortstypes). The pace of reading was also considerably slower.
|
Comment
The higher error rate in reading cannot be attributed to the fact that when not paced V.Y.G. reads at a fast pace, or to slowness of speech. In these samples, his spontaneous speech was faster than his reading. In conjunction with the findings of Experiment 1, these results add weight to the suggestion that the phonological word production deficit exhibited by V.Y.G. is restricted to his reading and does not impinge upon other modalities of word production.
|
Discussion |
|---|
|
TopAbstractIntroductionCase reportNeuropsychological backgroundDocumentation of reading skillsExperimental investigationsDiscussionAcknowledgementsReferences |
|---|
We have described the case of a patient whose reading is characterized by multiple phonological errors. This phonological word production deficit does not affect his naming or his spontaneous speech, which remains fluent and voluble. V.Y.G. shows significant effects of word frequency and regularity on reading tasks. We have also demonstrated that his reading is strongly influenced by temporal factors, in particular the RSI. This temporal condition effect was found to be restricted to reading tasks; the same effect was not observed on tests of naming or repetition. Tests of irregular word reading and comprehension conclusively demonstrate that he is able to comprehend words which he cannot read aloud.
In this discussion, we will consider two particular issues. First, we will discuss the implications of these findings for the dynamics of the word production system and the concept of refractoriness. Secondly, we will propose that the discrepancy between reading and speaking shown by V.Y.G. addresses the debate over multiple, task-dependent output systems in cognitive neuropsychological models of language production.
System dynamics and refractoriness
Our results demonstrate that V.Y.G. makes a high proportion of phonological errors on reading tasks which are not observed in other modalities of word production. However, we also show that these phonological errors are partially ameliorated by increasing the RSI interval. Even a modest increase in interval duration from 2 to 5 s was noted to result in significantly fewer phonological reading errors.
The facilitation of performance by increasing the RSI has been observed previously in cases of refractory access disorders, in which information in a system becomes temporarily unavailable following activation. These disorders represent one type of `access' disorder in which access to information is dependent upon temporal factors. Refractory access disorders have been described in word and picture comprehension (Warrington and McCarthy, 1983, 1987; Cipolotti and Warrington, 1995; McNeil et al., 1995) and, more recently, in picture naming (McCarthy and Kartsounis, 2000). To our knowledge, no previous case of a refractory deficit affecting word reading has been reported.
The reading deficit observed in V.Y.G. shares the temporal features of the previous cases of refractory access disorders. However, V.Y.G.'s comprehension of the written word remains relatively intact. Thus we would argue that his reading errors cannot be attributed to a semantic access deficit. Of the 199 reading error responses, only 19 were not classified as phonemic paraphasias; these comprised neologistic errors and letter substitution errors resulting in a real word. This refractory access deficit appears to be very localized in so far as we have only observed an effect of temporal factors in reading and indeed not in repetition, naming or in spontaneous speech, which is produced more quickly than his reading responses.
Our analysis of the reading errors made by V.Y.G. revealed that approximately two-thirds were phonemic paraphasias, involving a mixture of real word (e.g. valebale) and neologistic errors (e.g. mincekense). To account for a similar pattern of errors in a patient of his own, Lecours proposed a phonemic activation model to describe the phoneme selection process (Lecours, 1976). Individual phonemes and frequently occurring phoneme clusters were hypothesized to be represented as specific units within the system, organized in terms of their similarities. Representations at the single word level would be transcoded into phonemes by the activation of the appropriate phonemic units. Lecours suggested that phonemic paraphasic errors might result from abnormally brief or abnormally long-lasting activation of units in this system.
The disruption in activation hinted at by Lecours is in part similar to the `abnormal activation' explanation of verbal perseverations provided by Cohen and Dehaene (Cohen and Dehaene, 1998). These researchers claimed that if a processing system is deafferented from its input, remaining activation will fail to be overcome by current input and indeed will summate across consecutive trials. The remaining activation might consequently cause a previous response to be repeated, resulting in a perseverative error. Refractoriness could possibly be conceptualized as a temporary inability to activate a target processing unit because of extraneous activation from previous trials. However, Cohen and Dehaene predict that the accumulated activation which disrupts processing of new information will also cause perseverative errors. In the present case, we only found evidence of perseveration of the initial phoneme from one target to the subsequent target in a negligible number of cases. Therefore, the Cohen and Dehaene model does not appear to fit the pattern of reading errors observed in V.Y.G.
An alternative explanation of refractoriness without perseveration has been proposed to account for a patient with intact comprehension who was demonstrated to have a refractory anomia (F.A.S. in McCarthy and Kartsounis, 2000). In a series of repeated naming trials, F.A.S. showed improved naming performance when a 10 s RSI was established. Furthermore no perseverative responses were noted. In contrast, F.A.S. was able to read the same words under speeded conditions with no impairment. McCarthy and Kartsounis account for F.A.S.'s anomia by referring to models of speech production which suggest that activation processes for word selection are in a continual state of flux. They suggest that in F.A.S., this flux has become retarded, with the dominant candidate unit exerting lateral inhibition over its competitors for a prolonged period. This excessive inhibition is posited to result in a temporary obstruction of the selection process required to make an appropriate subsequent response.
Although V.Y.G. is held to show impairment at a different level of processing, namely the phoneme selection level, a similar amount of excessive lateral inhibition may well account for his phonemic paraphasic errors. The activation of a set of phoneme units required to read one word may result in excessive lateral inhibition of neighbouring units, hence temporarily blocking access to phoneme units required to read subsequent items.
In their account of F.A.S., McCarthy and Kartsounis also propose that refractory disorders may be seen as the complementary syndrome to the phenomenon of perseveration. Perseveration, they suggest, may be the consequence of an abnormal persistence of activation, whereas some refractory disorders may be the consequence of abnormally persistent inhibition. The present case provides an opportunity to examine the relationship between perseveration and refractoriness by comparing and contrasting V.Y.G.'s reading impairment with the case described as a fluent phonemic jargon dyslexic (H.P. in Warrington and Leff, 2000).
We shall focus here upon first the similarities and secondly the differences between their respective dyslexic syndromes. Both V.Y.G. and H.P. make a significant number of phonological errors when reading aloud. Direct comparison on the Patterson and Hodges corpus of monosyllabic words (Patterson and Hodges, 1992) shows a closely matched performance in terms of both percentage total error rate (V.Y.G., 44%; H.P., 43%) and percentage phonological error rate (V.Y.G., 63%; H.P., 71%). Both V.Y.G. and H.P. also demonstrate an ability to comprehend words which they cannot read aloud (V.Y.G., 24 out of 25 and 13 out of 25; H.P., 25 out of 25 and 14 out of 25 for comprehension and mean reading score of the Funnell irregular words, respectively). Consequently, the locus of the deficit in both patients is suggested to be at the phoneme selection level. V.Y.G. and H.P. also both showed inconsistent performance on three repeated reading trials of the Funnell words (2 = 0.10, P > 0.5 and 2 = 0.57, P > 0.1, respectively).
Two major differences distinguish the reading performance of V.Y.G. and H.P. First, H.P.'s performance was characterized by a high number of perseverations whereas V.Y.G.'s was not. On the Patterson and Hodges corpus of words, H.P.'s perseverative error rate was 44% whereas V.Y.G.'s was only 7%. H.P. also made a small number of whole word perseverative errors of which V.Y.G. showed no evidence (e.g. the sequence: dump–dump, comb–come, where–where, hoop–dump). Secondly, though both patients showed inconsistent performance when reading aloud, H.P. showed no evidence of sensitivity to temporal factors. Thus not all access disorder cases can be characterized as refractory, as response inconsistency may be seen in the absence of a sensitivity to temporal factors. However, V.Y.G.'s performance was facilitated by increasing the RSI. Here we would argue that our patient, V.Y.G., can be characterized as having a refractory access type of disorder.
H.P.'s dyslexic syndrome was also accounted for in terms of a failure, not an excess, of normal inhibitory processes (Warrington and Leff, 2000). They argued that the absence of temporal factors together with the high incidence of perseverative errors required an account in terms of too little inhibition rather than abnormal levels of activation (Cohen and Dehaene, 1998). According to Cohen and Dehaene, the cessation of input ought to allow any persistent activation to dissipate, but the introduction of a response–stimulus delay after each reading response did not reduce either H.P.'s overall number of errors or, more specifically, the number of perseverations he made. However, a different account is required for our patient V.Y.G. We would suggest that the refractory access disorder is accounted for by excess lateral inhibition at the phoneme selection level. This would account for V.Y.G.'s difficulty in producing an appropriate response and for the facilitation in performance caused by introducing a delay after each response.
Although we have suggested that one syndrome involves an excess of inhibition and the other a dearth of inhibition, the difference in the temporal dynamics of the two impaired systems remains salient. The excess lateral inhibition posited to explain V.Y.G.'s reading deficit is assumed to resolve after a short refractory period of relatively prescribed duration. This excess inhibition may reflect a shift in the balance of excitatory and inhibitory cells in local networks. Paradoxically, it may be that neuronal damage can shift the balance in the opposite direction, leaving a system with a temporary lack of inhibition. In H.P., the damage to the system is such that the return to a resting level of excitation following activation of a phoneme unit(s) is a non-refractory period of irregular duration.
Bringing the hypothesized mechanisms of refractoriness and perseveration together, it appears that the system characteristics of temporal sensitivity and perseveration may form a double dissociation. V.Y.G. demonstrates sensitivity to temporal factors but not perseveration. H.P. demonstrates perseveration but not sensitivity to temporal factors. Therefore, it appears that perseveration is not linked causally to the dynamics of the system. Rather, it is the direction of change in an inhibitory system (towards excess or failure) which determines whether access errors are perseverative or not. However, it is unclear at present whether temporal sensitivity and perseveration are mutually exclusive. If this were not the case, we might expect to see perseverative errors in reading or other cognitive domains which are sensitive to temporal factors. We are unaware of the documentation of any such cases. Either way, with the documentation of refractory disorders of word and picture comprehension, naming and now reading, there is growing confidence that any module of cognition may become refractory as well as unavailable.
Single or multiple output mechanisms?
We now turn to the issue of the discrepancy between the reading and spontaneous speech of V.Y.G. The prevalence of phonemic paraphasias in V.Y.G.'s reading has already been discussed. The fact that V.Y.G. can comprehend words which he cannot read aloud indicates that the locus of his dyslexic deficit lies at the level of a post-semantic phonological output store. In contrast, his spontaneous speech was found to be relatively free from such phonological errors. In a direct comparison, V.Y.G. was shown to be unable to read a transcript of his own speech, which in terms of phonology was accurate. This provides further evidence that the phonemic errors observed on reading tasks do not stem from the target words falling outside his vocabulary. This finding also raises questions regarding the architecture of the phonological output store itself. If a single multipurpose store is involved in the phonemic selection for all word production, why is V.Y.G.'s deficit task-specific to reading?
The suggestion that there may be separate output mechanisms for reading and naming was made by Orpwood and Warrington who described an anomic patient, M.R.F., who showed intact comprehension of both written and spoken words and virtually intact single word reading (Orpwood and Warrington, 1995). Furthermore, M.R.F. made semantic errors on naming and spelling but not reading tasks. Previous work had suggested that semantic errors may arise from damage at the lexical output level (Caramazza and Hillis, 1990). Given M.R.F.'s intact semantic knowledge of the words on which he made naming and spelling errors, Orpwood and Warrington claimed that M.R.F. had damage to the phonological and orthographic output stores, respectively. However, the observation that M.R.F. could read the names of objects he was unable to name led the researchers to propose an output store used specifically for reading. A model including separate output stores for reading and naming remains compatible with previous findings of concomitant oral reading and naming deficits (Caramazza and Hillis, 1990). The two spoken output processes might well be assumed to have anatomically close loci of representation.
An alternative explanation of this pattern of naming and reading deficits posits that `apparent dissociations between tasks are attributable to differential task demands' (Lambon-Ralph et al., 1999, p.166). Naming is suggested to be the modality most vulnerable to damage because the only source of phonological activation comes from the semantic system (Humphreys et al., 1995). Reading is viewed as more resilient as there are two sources of phonological activation—orthographic and semantic (Coltheart et al., 1992; Plaut et al., 1996). These two sources of activation are claimed to combine (or `summate': cf. Hillis and Caramazza, 1991, 1995), leaving the reading process less vulnerable to damage. The facilitation of naming performance with multiphonemic cueing supports the claim that different sources of phonological information can combine to activate word representations in the naming process (Lambon-Ralph et al., 1999). Furthermore, a variety of cueing procedures (e.g. phonemic, sentence frame, picture frame) have been shown to aid impaired naming performance (e.g. Williams and Canter, 1982, 1987; Breen and Warrington, 1995; Lambon-Ralph, 1998). The proposition of a task-specific output store does not assume different properties of the naming process. More importantly, Breen and Warrington found there was no facilitation from reading a word to naming it (Breen and Warrington, 1995). Lambon-Ralph and colleagues provide no account of why some forms of activation at the phonological output store combine to facilitate naming whilst others do not (Lambon-Ralph et al., 1999).
The `differential task demand' explanation of Lambon-Ralph and colleagues also makes a stronger prediction: that naming and reading will create a single and not a double dissociation. Two aspects of the present case cause us to challenge this claim. First, V.Y.G.'s naming performance is worse than his reading in terms of the total number of correct responses. However, the patterns of performance are very different. Phonological errors characterize V.Y.G.'s reading but are rarely observed in his naming. Although weak on naming tasks, V.Y.G.'s performance was not at floor as he was still able to name 12 out of 40 object pictures and 11 out of 40 action pictures. The impairment in naming performance in the absence of phonemic paraphasic errors and intact comprehension suggests that the locus of V.Y.G.'s naming deficit may be post-semantic and pre-phonological. If reading is more resistant to output level damage than naming, as Lambon-Ralph and colleagues suggest (Lambon-Ralph et al., 1999), damage to the stronger reading process should necessarily also affect the weaker naming ability. However, phonological naming errors are not observed in V.Y.G. The occurrence of phonological errors in reading together with the absence of such errors in naming provides evidence that there is indeed a double dissociation between between reading and naming, and that the more commonly observed relative sparing of reading cannot be attributed to differential task demands.
Secondly, the dissociations discussed so far are between reading and naming. In the present case, however, the most striking dissociation is that between reading and another oral output process: spontaneous speech. V.Y.G. shows a selective impairment for reading while his speech remains phonologically intact. Any explanation of V.Y.G.'s performance in terms of differential task demands does not seem appropriate as the reading process invokes phonological activation from at least as many sources as spontaneous speech. Given that our patient is held to exhibit damage at the phoneme selection level, we suggest that a double dissociation exists between the output stores required for reading and speaking. Whilst not referring to the exact dissociation discussed by Lambon-Ralph and colleagues (Lambon-Ralph et al., 1999), these findings appear difficult to reconcile with the concept of a single, multipurpose phonological output store.
|
Acknowledgements |
|---|
|
TopAbstractIntroductionCase reportNeuropsychological backgroundDocumentation of reading skillsExperimental investigationsDiscussionAcknowledgementsReferences |
|---|
We wish to thank Dr Nick Fox and Dr John Stevens for their assessment of the neuroimaging data, and Dr Lisa Cipolotti for making available neuropsychological test results. We are particularly grateful to Professor Martin Rossor for making the original observation that led to this investigation and for encouraging us to carry out this single case study of a patient under his care.
|
References |
|---|
|
TopAbstractIntroductionCase reportNeuropsychological backgroundDocumentation of reading skillsExperimental investigationsDiscussionAcknowledgementsReferences |
|---|
Baxter DM, Warrington EK. Category specific phonological dysgraphia. Neuropsychologia 1985; 23: 653–66.[Web of Science][Medline]
Baxter DM, Warrington EK. Measuring dysgraphia: a graded-difficulty spelling test. Behav Neurol 1994; 7: 107–16.
Breen K, Warrington EK. Impaired naming and preserved reading: a complete dissociation. Cortex 1995; 31: 583–8.[Web of Science][Medline]
Burgess PW, Shallice T. Bizarre responses, rule detection and frontal lobe lesions. Cortex 1996; 32: 241–59.[Web of Science][Medline]
Caramazza A, Hillis AE. Where do semantic errors come from? Cortex 1990; 26: 95–122.[Web of Science][Medline]
Caramazza A, Hillis AE. Lexical organization of nouns and verbs in the brain. Nature 1991; 349: 788–90.[Medline]
Cipolotti L, Warrington EK. Towards a unitary account of access dysphasia: a single case study. Memory 1995; 3: 309–32.[Web of Science][Medline]
Cohen L, Dehaene S. Competition between past and present: assessment and interpretation of verbal perseverations. Brain 1998; 121: 1641–59.
Coltheart M. Reading, phonological recoding and deep dyslexia. In: Coltheart M, Patterson K, Marshall JC, editors. Deep dyslexia. London: Routledge; 1980. p. 197–226.
Coughlan AK, Warrington EK. Word-comprehension and word-retrieval in patients with localized cerebral lesions. Brain 1978; 101: 163–85.
Funnell E. Phonological processes in reading: new evidence from acquired dyslexia. Br J Psychol 1983; 74: 159–80.[Web of Science]
Glushko RJ. The organization and activation of orthographic knowledge in reading aloud. J Exp Psychol Hum Percept Perform 1979; 5: 674–91.
Hillis AE, Caramazza A. Mechanisms for accessing lexical representations for output: evidence from a category-specific semantic deficit. Brain Lang 1991; 40: 106–44.[Web of Science][Medline]
Hillis AE, Caramazza A. Converging evidence for the interaction of semantic and sublexical phonological information in accessing lexical representations for spoken output. Cognit Neuropsychol 1995; 12: 187–227.
Jackson M, Warrington EK. Arithmetic skills in patients with unilateral cerebral lesions. Cortex 1986; 22: 611–20.[Web of Science][Medline]
Lambon-Ralph MA. Distributed versus localist representations: evidence from a study of item consistency in a case of classical anomia. Brain Lang 1998; 64: 339–60.[Web of Science][Medline]
Lambon Ralph MA, Cipolotti L, Patterson K. Oral naming and oral reading: do they speak the same language? Cognit Neuropsychol 1999; 16: 157–69.
Lecours AR. The `pure form' of the phonetic disintegration syndrome (pure anarthria): anatomo-clinical report of a historical case. Brain Lang 1976; 3: 88–113.[Web of Science][Medline]
McCarthy RA, Kartsounis LD. Wobbly words: refractory anomia with preserved semantics. Neurocase 2000; 6: 487–97.[Web of Science]
McCarthy RA, Warrington EK. A two route model of speech production: evidence from aphasia. Brain 1984; 107: 463–85.
McKenna P, Warrington EK. The Graded Naming Test. Windsor (UK): NFER-Nelson; 1983.
McNeil JE, Cipolotti L, Warrington EK. The accessibility of proper names. Neuropsychologia 1994; 32: 193–208.[Web of Science][Medline]
Morton J. Facilitation in word recognition experiments causing changes in the logogen model. In: Kolers PA, Wrolstad ME, Bouma H, editors. Processing of visible language, Vol. 1. New York: Plenum Press; 1979.
Morton J. The logogen model and orthographic structure. In: Frith U, editor. Cognitive approaches to spelling. London: Academic Press; 1980. p. 117–34.
Morton J, Patterson KE. A new attempt at an interpretation, or, an attempt at a new interpretation. In: Coltheart M, Patterson K, Marshall JC, editors. Deep dyslexia. London: Routledge; 1980. p. 91–118.
Neary D, Snowden JS, Gustafson L, Passant U, Stuss D, Black S, et al. Frontotemporal lobar degeneration: a consensus on clinical diagnostic criteria. [Review]. Neurology 1998; 51: 1546–54.
Nelson HE. National Adult Reading Test (NART): for the assessment of premorbid intelligence in patients with dementia. Windsor (UK): NFER-Nelson; 1982.
Orpwood L, Warrington EK. Word specific impairments in naming and spelling but not reading. Cortex 1995; 31: 239–65.[Web of Science][Medline]
Patterson K, Hodges JR. Deterioration of word meaning: implications for reading. Neuropsychologia 1992; 30: 1025–40.[Web of Science][Medline]
Plant DC, McClelland JL, Seidenberg MS, Patterson K. Understanding normal and impaired word reading: computational principles in quasi-regular domain. Psychol Rev 1996; 103: 56–115.[Web of Science][Medline]
Warrington EK. Recognition Memory Test. Windsor (UK): NFER-Nelson; 1984.
Warrington EK, Cipolotti L. Word comprehension: the distinction between refractory and storage impairments. Brain 1996; 119: 611–25.
Warrington EK, Leff A. Jargon dyslexia: a single case study of intact reading comprehension in a jargon dysphasic. Neurocase 2000; 6: 499–507.[Web of Science]
Warrington EK, McCarthy R. Category specific access dysphasia. Brain 1983; 106: 859–78.
Warrington EK, McCarthy RA. Categories of knowledge. Brain 1987; 110: 1273–96.
Warrington EK, Shallice T. Semantic access dyslexia. Brain 1979; 102: 43–63.
Warrington EK, James M, Maciejewski C. The WAIS as a lateralizing and localizing diagnostic instrument: a study of 656 patients with unilateral cerebral lesions. Neuropsychologia 1986; 24: 223–39.[Web of Science][Medline]
Weigl E. Zur Psychologie sogenannter Abstraktionsprozesse. Z Psychol 1927; 103: 2–45. [Translated by Rioch M and reprinted. On the psychology of so called processes of abstraction. J Abnorm Soc Psychol 1948; 36: 3–33.]
Williams S, Canter G. The influence of situational context on naming performance in aphasic syndromes. Brain Lang 1982; 6: 63–74.
Williams S, Canter G. Action naming performance in four syndromes of aphasia. Brain Lang 1987; 32: 124–36.[Web of Science][Medline]
Received December 21, 2000. Revised March 29, 2001. Accepted April 2, 2001.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  S. J. Crutch and E. K. Warrington Abstract and concrete concepts have structurally different representational frameworks Brain, March 1, 2005; 128(3): 615 - 627. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||