Nesropsychologra Vol 30,No F’nnted III Great Bntan
12 pp 1025-1040
DETERIORATION
1992
OF WORD MEANING: READING
KARALYN
PATTERSON*
MRC’ Applied Psychology Unit, Cambrldgc; Addenbrookc’s
IMPLICATIONS
FOR
and JOHN R. Houc;t-.s and University of Cambridge Hospital. U.K.
Clinical School.
Abstract---We investigated six patients with progressive focal dementia or progressive aphasia. who showed impairments in knowlcdgc of word meaning ranging from moderate to very severe. In all GISCS,a test of oral word reading demonstrated preswed reading of \+ords with regular spelling-tosound correspondences (e.g. MINT). but Impaired wading of words uith atypical correspondences (e.g. PINT). The lcvcl of success on these “exception” words was signilicantly related to word frcqucncy. and the most common error was the assignment of a more typical spelling-sound corrcapondence. Various explanations are considered for this common association between loss of word meaning and a wrface alcxic pattern of reading performance.
INTRODUCTION NEUROLOGICAL DISEASE or injury may disrupt a person’s ability to understand the meaning of spoken and written words. The question addressed in this paper is whether such disruption of semantic memory or word meaning has predictable consequences for components of reading skill. The conclusion offered here is that deterioration of word meaning leads to a particular form of reading disorder designated su~fircr crle.uia or SW$KC rl~tdcsia 130, 351, also sometimes called .selec.ticr pwscwatior~ (!/‘phorlologic,a/ wadirzy [44]. In its purest form, surface alexia (in English) can be considered to have three main characteristics (see Ref. 141): (I) normal or near-normal reading aloud, as regards both accuracy and speed. of words whose pronunciations are predictable from their spelling patterns, that is, words with a regular spelling-to-sound correspondence (like MINT); (2) significantly reduced accuracy in reading of letter strings whose pronunciations cannot be so predicted, that is words with an exceptional spelling-to-sound correspondence (like PINT); and (3) errors in reading cxccption words that are predominantly “pure regularizations” (pronouncing PINT to rhyme with HINT, PRINT, LINT, etc.). Disorders of semantic memory are typically associated with impaired object naming [16, 551. Because the relationship between the visual appearance of an object and its phonological label is purely arbitrary, a person can only name an object that is already known: there would be no basis for guessing or inferring the phonological label for an unfamiliar object. Therefore, the process of object naming probably relies on some form of semantic mediation between object perception and phonological retrieval. Although some authors have suggested that naming of familiar objects could in principle reflect direct.
*Address for correapondcncc: (‘HZ 2Eb-. U.K.
K. E. Patterson.
MRC
Applied
Psychology Unit.
I5 Chaucer
Road, Cambridge
1026
K. Pi\r-I~KSO\ and J.R. HOIXM
non-semantically mediated, associations between visual object representations and names [ 13, 22, 391, the pervasive neuropsychological association between impaired comprehension and impaired naming has led most neuropsychologists to adopt the position that naming requires intact knowledge of the object’s meaning or function 117, 191. Words written in alphabetic script have a principled rather than an arbitrary relationship between visual form and phonological label; thus, in distinction to an unfamiliar object, an unfamiliar string of letters does afford computation of a phonological representation. There are however few writing systems where general knowledge of the regularities of spelling-to-sound correspondence would yield the conventionally correct pronunciation (supra-segmental as well as segmental) of all lexical entries; and of course in a language like English, with its notorious inconsistencies cvcn at the segmental level, the reader needs explicit instruction on words with an exceptional spelling-to-sound correspondence (such as PINT or GAUGE) to pronounce them correctly. Of such words, one can therefore ask a question similar to that posed above for object naming. Does successful oral reading of familiar exception words reflect the role of word comprehension in reading‘? Or has the reader learned to rctricvc or compute the word’s pronunciation directly from the orthographic input? This question can be addressed with various sources of evidence, including relevant neuropsychological data. Suppose that a patient with a comprehension disorder fails to understand the meaning of an exception word like PINT (but retains components of reading skill such as letter recognition and knowledge of letter sounds). If that patient is asked to read aloud PINT and other exception words, can one predict whether the patient will pronounce these words correctly, or show a surface alexic pattern of reading with regularization errors? The available neuropsychological evidence provides a less than straightforward answer to this question. Although there have been a number of reports of surface alexia in patients with impairments of word comprehension [ 1. 24, 34. 551, there is also a well documented exception to this association in an important longitudinal study by SCHWARTZ rt (I/. [40]. This patient, W.L.P., did in fact show some surface alexic characteristics at a later stage of her progressive dementing disease: but there was a stage at which, though W.L.P.‘s her exception word reading was uncomprehension had significantly deteriorated. impaired. This observed dissociation argues against the notion that an intact semantic representation is required as mediation between visual word analysis and assignment of correct pronunciation to exception words. The present paper will provide new neuropsychological data on the relationship between semantic memory and reading. This report has a somewhat unusual format: it is neither a group study, nor a typical single-case study, nor even a scrics of complctc singlecase studies. Rather it represents a focus on one specific aspect of six single-case studies. Each of these patients has a demonstrable deterioration of word meaning, ranging from of progrcssivc brain disease. The most moderate to profound. as a consequence informative description of these cases is a pattern of focal semantic dementia [16. 4X] or progressive aphasia 131. 32, 37. 541. Only brief case histories will be prcscnted here. Two of the six cases have previously been described in the literature: the remaining four arc part of a longitudinal study of and will eventually be described in that context. The semantic memory deterioration. striking pattern of performance by these six patients on a specific test of word reading provides the basis for this report.
DETERIORATIONOF WORD MEANING: IMPLICATIONSFOR READING
1027
CASE REPORTS The case histories will be organized into two groups corresponding to a rough classification of severity of deficit in vocabulary and word comprehension. These assignments were made on the basis of a variety of measures rather than any single test, partly because exactly comparable test data are not available for all six cases, but also because of our conviction that no single test adequately captures this ability. The quality of the patients’ spontaneous speech has been a significant factor in this classification: since all six patients have adequate phonological production (in the sense that all have intact single-word repetition), and since none of the patients is agrammatic, we assume that the main limiting factor in their spontaneous speech is vocabulary loss. A sample ofeach patient’s speech is included in the case reports below. Naming is another crucial and sensitive indicator of vocabulary loss. All patients were given various picture naming tests, including (for five of the six cases) a 48-item subset of SNODGRASS and VAIZDERWART 1471pictures used by HOLXXS et ul. [14, 161. These can be divided into 24 higher and 24 lower frequency names for an assessment of frequency effects on object naming, which is particularly germane to a comparison of the patients’ naming and reading abilities.
K.T.: M&ARTHY and WARRINGTON [24].K.T., a right-handed 54-year-old businessman, was investigated for possible pre-senile dementia following a 4-year deterioration in social and cognitive abilities, especially speech comprehension. He received average or dull average WAIS scores on Digit Symbol, Block Design and Digit Span, but was unable to score on either Vocabulary or Similarities. He failed to reach baseline on the Peabody Picture Vocabulary Test, whether the words were given in spoken or written form, consistently responding “I do not know the word”. In a worddpicture matching test (one spoken word, five pictures) with the 130 higher frequency items from SNODGRASS and VANUERWART [47], K.T.‘s score was only slightly above the level of chance (38 and 34% correct, on two different occasions). He managed to name I1 % of these high-frequency Snodgrass and Vanderwart pictures, but failed to score at all on two other naming tests 127, 337. Although this is the one case, of the six considered here, whose pattern of performance might indicate a generalized dementia, McCarthy and Warrington report that K.T.‘s perceptual and spatial functions were retained at a very high level; this fact argues against a diagnosis of dementia of Alzheimer’s type (DAT). Spontaneous speech: (Describing his problems) “This is the problem also, I used to understand. All the Indian words, I’ve forgotten it. But I used to, because I was, on the film. that’s why I knew all the words, but I’ve forgotten all that, I can’t understand any of the these words.” P.P.: Houcazs and PATTERSON [I51.P.P., a 69.year-old right-handed woman who had worked as a clerical olhcer and secretary to a general practitioner, presented with a 3-year history of progressive loss of vocabulary and comprehension. Positron emission tomography scanning revealed marked reduction of cerebral perfusion and metabolism in the left temporo-parietal region. The tentative diagnosis is focal “semantic dementia”. She had above average non-verbal intelligence (Ravens Progressive Matrices), and obtained a dramatic range of WAIS sub-test age scaled scores from 17 (Digit Span) and I2 (Block Design) to 1 (Information) and 0 (Similarities). She was consistently at or near chance level on matching spoken words to pictures (LUVS from KAY et al. 1211, with live alternative pictures per word: 35% correct; a within-category test from HOIXES ct al. [16], with six alternatives per word: 29% correct). She was similarly at or near chance when asked yessno questions about either superordinate category membership (e.g. Is an apple a fruit’?: 65% correct) or features of objects (e.g. Does a zebra have stripes’?: 54%). P.P.‘s spontaneous speech was fluent and syntactically well formed but almost wholly lacking in specific content words. She has named no objects correctly on any test thus far administered (e.g. scores of zero on both the Boston Naming Test and the 48-item test of Hot)C;rs et al. [ 161): all responses were of the form “I wish I knew what it was”, or “I’m terrible at this”. She was also unable to offer even a single item in category verbal fluency tests. responding to the animal category for example with “I wish I could remember what an animal was”. Spontaneous speech: (from the BDAT cookie theft description) “He’s getting something from up there. And they’re washing, going down there. I don’t understand very much.” [Examiner, pointing to spilling water: And what’s happening here?] “It’s comingdown, isn’t it?” [Examiner, pointing to children: And what are they doing over here?] “Getting something down from there, or putting it up.” [Examiner: What are they getting down?] “I can’t read, I’m afraid. I don’t know what they’re getting down.”
Putients
with u more modrrate
loss of comprehension
T.O.B., a right-handed 64-year-old male civil servant, suffered a progressive deterioration in vocabulary and speech comprehension. CT scan was normal, but coronal slices from MRI revealed left perisylvian atrophy, and PET-scanning showed significantly reduced oxygen metabolism in the whole left temporal lobe and the left third frontal gyrua (TYRRELL et ul. [54]). With a performance IQ of 116 and a verbal IQ of 95, the main feature of note was a ‘I. total failure to define a number of relatively common words which must have once been well within his vocabulary” (MCCARTHY and WARRINGTON 1261,p. 429).For example, asked to describe a rhinoceros, he replied, “Animal, can’t give you any functions”; dolphin: ‘%a fish or a bird”. His abilities in confrontation picture naming were poor but not abolished: McCarthy & Warrington report that T.O.B. failed to T.O.B.
126.34.541.
score on M(.Kt.ti~ xnd WAKKI~~GTON‘Sobject naming test [77]. hut hc correctly named I7,‘JX = 3S”i, of the plcturc\ used by Ho~x;t,s (“I tri. [ 161 (Parkm, personal communication). lfthcsc 4X items UC dlvlded into the 24 higher ;Ind 24 lower frequency object names. T.O.B. correctly named I7 of the former but only 5 of the latter, 7’ (I P=O.O3.PAKIing all the thing\ hclorc hc +ot to what I really tianted. I mean. lfhc \+:Is golng out to get the shopping v. hilo I’m xt work, hc nwd\ to knou LIh;~t I( is to write it down. dwsn’t hc umm Those littlc white. no, not white grew. little gccn round things. umm .” ./.L. J.L. is ;L60.year-old ripht-handed man, formerly the poncral manager ofweral comp;mle\. In early I YYI hc prcscnted with a I?-month history of diminished ~ocnhulary. prohlcms in t-cmcmbcrinp the nxmch of people and place\, and difficulty in understanding cwn common words such ;I\ the n;imc\ of foods. CT and M R I vxnning dcmonstruted striking hllaterai temporal-lohc atrophy. HI\ episodic mcmorq was relolivclq intact: for excamplc. hc wxuratelq rclatcd dctaila of hi5 life ~wd hl\ work Spontwneous speech wits Ilucnt \\ith norm;il articulation :ind \yntnx. hut some word-finding difficulties. Good .\yntactic bkills were conlii-mcd by his pcrformancc on the Token Test (34. 361 :tnd good pcrccptual shills h) hi> copy thecomplex Rcy ligurc (33 361. By contrast. hc h;~d impalrccl knuwlcdgr
12 pp 1025-1040
DETERIORATION
1992
OF WORD MEANING: READING
KARALYN
PATTERSON*
MRC’ Applied Psychology Unit, Cambrldgc; Addenbrookc’s
IMPLICATIONS
FOR
and JOHN R. Houc;t-.s and University of Cambridge Hospital. U.K.
Clinical School.
Abstract---We investigated six patients with progressive focal dementia or progressive aphasia. who showed impairments in knowlcdgc of word meaning ranging from moderate to very severe. In all GISCS,a test of oral word reading demonstrated preswed reading of \+ords with regular spelling-tosound correspondences (e.g. MINT). but Impaired wading of words uith atypical correspondences (e.g. PINT). The lcvcl of success on these “exception” words was signilicantly related to word frcqucncy. and the most common error was the assignment of a more typical spelling-sound corrcapondence. Various explanations are considered for this common association between loss of word meaning and a wrface alcxic pattern of reading performance.
INTRODUCTION NEUROLOGICAL DISEASE or injury may disrupt a person’s ability to understand the meaning of spoken and written words. The question addressed in this paper is whether such disruption of semantic memory or word meaning has predictable consequences for components of reading skill. The conclusion offered here is that deterioration of word meaning leads to a particular form of reading disorder designated su~fircr crle.uia or SW$KC rl~tdcsia 130, 351, also sometimes called .selec.ticr pwscwatior~ (!/‘phorlologic,a/ wadirzy [44]. In its purest form, surface alexia (in English) can be considered to have three main characteristics (see Ref. 141): (I) normal or near-normal reading aloud, as regards both accuracy and speed. of words whose pronunciations are predictable from their spelling patterns, that is, words with a regular spelling-to-sound correspondence (like MINT); (2) significantly reduced accuracy in reading of letter strings whose pronunciations cannot be so predicted, that is words with an exceptional spelling-to-sound correspondence (like PINT); and (3) errors in reading cxccption words that are predominantly “pure regularizations” (pronouncing PINT to rhyme with HINT, PRINT, LINT, etc.). Disorders of semantic memory are typically associated with impaired object naming [16, 551. Because the relationship between the visual appearance of an object and its phonological label is purely arbitrary, a person can only name an object that is already known: there would be no basis for guessing or inferring the phonological label for an unfamiliar object. Therefore, the process of object naming probably relies on some form of semantic mediation between object perception and phonological retrieval. Although some authors have suggested that naming of familiar objects could in principle reflect direct.
*Address for correapondcncc: (‘HZ 2Eb-. U.K.
K. E. Patterson.
MRC
Applied
Psychology Unit.
I5 Chaucer
Road, Cambridge
1026
K. Pi\r-I~KSO\ and J.R. HOIXM
non-semantically mediated, associations between visual object representations and names [ 13, 22, 391, the pervasive neuropsychological association between impaired comprehension and impaired naming has led most neuropsychologists to adopt the position that naming requires intact knowledge of the object’s meaning or function 117, 191. Words written in alphabetic script have a principled rather than an arbitrary relationship between visual form and phonological label; thus, in distinction to an unfamiliar object, an unfamiliar string of letters does afford computation of a phonological representation. There are however few writing systems where general knowledge of the regularities of spelling-to-sound correspondence would yield the conventionally correct pronunciation (supra-segmental as well as segmental) of all lexical entries; and of course in a language like English, with its notorious inconsistencies cvcn at the segmental level, the reader needs explicit instruction on words with an exceptional spelling-to-sound correspondence (such as PINT or GAUGE) to pronounce them correctly. Of such words, one can therefore ask a question similar to that posed above for object naming. Does successful oral reading of familiar exception words reflect the role of word comprehension in reading‘? Or has the reader learned to rctricvc or compute the word’s pronunciation directly from the orthographic input? This question can be addressed with various sources of evidence, including relevant neuropsychological data. Suppose that a patient with a comprehension disorder fails to understand the meaning of an exception word like PINT (but retains components of reading skill such as letter recognition and knowledge of letter sounds). If that patient is asked to read aloud PINT and other exception words, can one predict whether the patient will pronounce these words correctly, or show a surface alexic pattern of reading with regularization errors? The available neuropsychological evidence provides a less than straightforward answer to this question. Although there have been a number of reports of surface alexia in patients with impairments of word comprehension [ 1. 24, 34. 551, there is also a well documented exception to this association in an important longitudinal study by SCHWARTZ rt (I/. [40]. This patient, W.L.P., did in fact show some surface alexic characteristics at a later stage of her progressive dementing disease: but there was a stage at which, though W.L.P.‘s her exception word reading was uncomprehension had significantly deteriorated. impaired. This observed dissociation argues against the notion that an intact semantic representation is required as mediation between visual word analysis and assignment of correct pronunciation to exception words. The present paper will provide new neuropsychological data on the relationship between semantic memory and reading. This report has a somewhat unusual format: it is neither a group study, nor a typical single-case study, nor even a scrics of complctc singlecase studies. Rather it represents a focus on one specific aspect of six single-case studies. Each of these patients has a demonstrable deterioration of word meaning, ranging from of progrcssivc brain disease. The most moderate to profound. as a consequence informative description of these cases is a pattern of focal semantic dementia [16. 4X] or progressive aphasia 131. 32, 37. 541. Only brief case histories will be prcscnted here. Two of the six cases have previously been described in the literature: the remaining four arc part of a longitudinal study of and will eventually be described in that context. The semantic memory deterioration. striking pattern of performance by these six patients on a specific test of word reading provides the basis for this report.
DETERIORATIONOF WORD MEANING: IMPLICATIONSFOR READING
1027
CASE REPORTS The case histories will be organized into two groups corresponding to a rough classification of severity of deficit in vocabulary and word comprehension. These assignments were made on the basis of a variety of measures rather than any single test, partly because exactly comparable test data are not available for all six cases, but also because of our conviction that no single test adequately captures this ability. The quality of the patients’ spontaneous speech has been a significant factor in this classification: since all six patients have adequate phonological production (in the sense that all have intact single-word repetition), and since none of the patients is agrammatic, we assume that the main limiting factor in their spontaneous speech is vocabulary loss. A sample ofeach patient’s speech is included in the case reports below. Naming is another crucial and sensitive indicator of vocabulary loss. All patients were given various picture naming tests, including (for five of the six cases) a 48-item subset of SNODGRASS and VAIZDERWART 1471pictures used by HOLXXS et ul. [14, 161. These can be divided into 24 higher and 24 lower frequency names for an assessment of frequency effects on object naming, which is particularly germane to a comparison of the patients’ naming and reading abilities.
K.T.: M&ARTHY and WARRINGTON [24].K.T., a right-handed 54-year-old businessman, was investigated for possible pre-senile dementia following a 4-year deterioration in social and cognitive abilities, especially speech comprehension. He received average or dull average WAIS scores on Digit Symbol, Block Design and Digit Span, but was unable to score on either Vocabulary or Similarities. He failed to reach baseline on the Peabody Picture Vocabulary Test, whether the words were given in spoken or written form, consistently responding “I do not know the word”. In a worddpicture matching test (one spoken word, five pictures) with the 130 higher frequency items from SNODGRASS and VANUERWART [47], K.T.‘s score was only slightly above the level of chance (38 and 34% correct, on two different occasions). He managed to name I1 % of these high-frequency Snodgrass and Vanderwart pictures, but failed to score at all on two other naming tests 127, 337. Although this is the one case, of the six considered here, whose pattern of performance might indicate a generalized dementia, McCarthy and Warrington report that K.T.‘s perceptual and spatial functions were retained at a very high level; this fact argues against a diagnosis of dementia of Alzheimer’s type (DAT). Spontaneous speech: (Describing his problems) “This is the problem also, I used to understand. All the Indian words, I’ve forgotten it. But I used to, because I was, on the film. that’s why I knew all the words, but I’ve forgotten all that, I can’t understand any of the these words.” P.P.: Houcazs and PATTERSON [I51.P.P., a 69.year-old right-handed woman who had worked as a clerical olhcer and secretary to a general practitioner, presented with a 3-year history of progressive loss of vocabulary and comprehension. Positron emission tomography scanning revealed marked reduction of cerebral perfusion and metabolism in the left temporo-parietal region. The tentative diagnosis is focal “semantic dementia”. She had above average non-verbal intelligence (Ravens Progressive Matrices), and obtained a dramatic range of WAIS sub-test age scaled scores from 17 (Digit Span) and I2 (Block Design) to 1 (Information) and 0 (Similarities). She was consistently at or near chance level on matching spoken words to pictures (LUVS from KAY et al. 1211, with live alternative pictures per word: 35% correct; a within-category test from HOIXES ct al. [16], with six alternatives per word: 29% correct). She was similarly at or near chance when asked yessno questions about either superordinate category membership (e.g. Is an apple a fruit’?: 65% correct) or features of objects (e.g. Does a zebra have stripes’?: 54%). P.P.‘s spontaneous speech was fluent and syntactically well formed but almost wholly lacking in specific content words. She has named no objects correctly on any test thus far administered (e.g. scores of zero on both the Boston Naming Test and the 48-item test of Hot)C;rs et al. [ 161): all responses were of the form “I wish I knew what it was”, or “I’m terrible at this”. She was also unable to offer even a single item in category verbal fluency tests. responding to the animal category for example with “I wish I could remember what an animal was”. Spontaneous speech: (from the BDAT cookie theft description) “He’s getting something from up there. And they’re washing, going down there. I don’t understand very much.” [Examiner, pointing to spilling water: And what’s happening here?] “It’s comingdown, isn’t it?” [Examiner, pointing to children: And what are they doing over here?] “Getting something down from there, or putting it up.” [Examiner: What are they getting down?] “I can’t read, I’m afraid. I don’t know what they’re getting down.”
Putients
with u more modrrate
loss of comprehension
T.O.B., a right-handed 64-year-old male civil servant, suffered a progressive deterioration in vocabulary and speech comprehension. CT scan was normal, but coronal slices from MRI revealed left perisylvian atrophy, and PET-scanning showed significantly reduced oxygen metabolism in the whole left temporal lobe and the left third frontal gyrua (TYRRELL et ul. [54]). With a performance IQ of 116 and a verbal IQ of 95, the main feature of note was a ‘I. total failure to define a number of relatively common words which must have once been well within his vocabulary” (MCCARTHY and WARRINGTON 1261,p. 429).For example, asked to describe a rhinoceros, he replied, “Animal, can’t give you any functions”; dolphin: ‘%a fish or a bird”. His abilities in confrontation picture naming were poor but not abolished: McCarthy & Warrington report that T.O.B. failed to T.O.B.
126.34.541.
score on M(.Kt.ti~ xnd WAKKI~~GTON‘Sobject naming test [77]. hut hc correctly named I7,‘JX = 3S”i, of the plcturc\ used by Ho~x;t,s (“I tri. [ 161 (Parkm, personal communication). lfthcsc 4X items UC dlvlded into the 24 higher ;Ind 24 lower frequency object names. T.O.B. correctly named I7 of the former but only 5 of the latter, 7’ (I P=O.O3.PAKIing all the thing\ hclorc hc +ot to what I really tianted. I mean. lfhc \+:Is golng out to get the shopping v. hilo I’m xt work, hc nwd\ to knou LIh;~t I( is to write it down. dwsn’t hc umm Those littlc white. no, not white grew. little gccn round things. umm .” ./.L. J.L. is ;L60.year-old ripht-handed man, formerly the poncral manager ofweral comp;mle\. In early I YYI hc prcscnted with a I?-month history of diminished ~ocnhulary. prohlcms in t-cmcmbcrinp the nxmch of people and place\, and difficulty in understanding cwn common words such ;I\ the n;imc\ of foods. CT and M R I vxnning dcmonstruted striking hllaterai temporal-lohc atrophy. HI\ episodic mcmorq was relolivclq intact: for excamplc. hc wxuratelq rclatcd dctaila of hi5 life ~wd hl\ work Spontwneous speech wits Ilucnt \\ith norm;il articulation :ind \yntnx. hut some word-finding difficulties. Good .\yntactic bkills were conlii-mcd by his pcrformancc on the Token Test (34. 361 :tnd good pcrccptual shills h) hi> copy thecomplex Rcy ligurc (33 361. By contrast. hc h;~d impalrccl knuwlcdgr
