The emergence of jargon in progressive fluent

Cognitive Neuropsychology

ISSN: 0264-3294 (Print) 1464-0627 (Online) Journal homepage: http://www.tandfonline.com/loi/pcgn20

The emergence of jargon in progressive fluent dysgraphia: The widening gap between target and response Naida L. Graham , Karalyn Patterson & John R. Hodges To cite this article: Naida L. Graham , Karalyn Patterson & John R. Hodges (2001) The emergence of jargon in progressive fluent dysgraphia: The widening gap between target and response, Cognitive Neuropsychology, 18:4, 343-361 To link to this article: http://dx.doi.org/10.1080/02643290125983

Published online: 09 Sep 2010.

Submit your article to this journal

Article views: 32

View related articles

Citing articles: 13 View citing articles

Full Terms & Conditions of access and use can be found at http://www.tandfonline.com/action/journalInformation?journalCode=pcgn20 Download by: [Mrc Cognition Brain Sci Unit]

Date: 25 May 2016, At: 07:44

COGNITIVE NEUROPSY CHOLOGY, 2001, 18 (4), 343–361

THE EMERGENCE OF JARGON IN PROGRESSIVE FLUENT DYSGRAPHIA: THE WIDENING GAP BETWEEN TARGET AND RESPONSE Naida L. Graham University of Cambridge and MRC Cognition and Brain Sciences Unit, Cambridge, UK

Downloaded by [Mrc Cognition Brain Sci Unit] at 07:44 25 May 2016

Karalyn Patterson MRC Cognition and Brain Sciences Unit, Cambridge, UK

John R. Hodges University of Cambridge and MRC Cognition and Brain Sciences Unit, Cambridge, UK

This paper presents a long-term follow-up study of a dysgraphic patient reported in an earlier paper (Graham, Patterson, & Hodges, 1997). FM developed “jargon dysgraphia”: In writing to dictation, she fluently produced well-formed written output consisting of pronounceable neologisms which, over time, bore a progressively weaker resemblance to the target words (e.g., scribe ® SCRIPE, January 1992; ® SRIME, November, 1994; ® BRINCE, August, 1996; ® MEATH, November, 1997). Although this spelling disorder has been (rarely) reported in patients with stable brain lesions, this is the first time that it has been documented in the context of progressive disease, a situation that provided a unique opportunity to document its characteristics at different stages. Longitudinal results from a spelling-to-dictation task showed that FM was, for a time, able to activate at least partial orthographic information, even after her overall accuracy (assessed in terms of the number of words spelled completely correctly) had reached the floor. For example, her error responses were correlated in length with the targets, and contained a larger than chance number of letters that overlapped between target and response. Ultimately, FM’s spelling responses had no detectable similarity with the targets, although they remained word-like in that they were pronounceable and contained few illegal letter combinations.

The term “jargon dysgraphia” has been used to describe a number of different writing impairments, all of which have in common with each other, and with jargon aphasia, fluent yet incorrect and often perseverative production of paragraphias, and neologisms that follow the graphotactic rules of the language. Jargon dysgraphia is thought to be rare, and has most often been described in the context of crossed aphasia (language impairment caused by damage to the nondominant hemisphere,

usually right-sided damage in a right-hander) (Alexander, Fischette, & Fischer, 1989; Assal, 1982; Bramwell, 1899; Demeurisse, Hublet, Coekaerts, Derouck, & Capon, 1986; Habib, Joanette, Ali-Cherif, & Poncet, 1983; Hashimoto, Tanaka, & Yoshida, 1998; Ihori, Kashiwagi, Kashiwagi, & Tanabe, 1994; Pillon, Desi, & Lhermitte, 1979; Washimi, Makishita, Miyasaka, Fujita, & Yanagisawa, 1987; Yokoyama, Okubo, Doseki, & Yamadori, 1981), although it has also

Requests for reprints should be addressed to Professor John R. Hodges, MRC-CBU, 15 Chaucer Road, Cambridge CB2 2EF, UK (Tel: 01223 355294; Email: [email protected]). Ó 2001 Psychology Press Ltd http://www.tandf.co.uk/journals/pp/02643294.html

343

DOI:10.1080/02643290042000161


GRAHAM, PATTERSON, HODGES

been reported in conjunction with aphasia from more standard left-hemisphere lesions (Cappa, Cavallotti, & Vignolo, 1987; Kertesz & Benson, 1970; Kinsbourne & Warrington, 1963; Lecours & Rouillon, 1976; Schonauer & Denes, 1994). Jargon dysgraphia characterised by fluent but neologistic spelling, which is the focus of this report, has mainly been described clinically (e.g., Cases I, II, V, & VI, Basso, Capitani, Laiacona, & Zanobio, 1985; Basso, Taborelli, & Vignolo, 1978; Cases 1, 3, 5, & 6, Cappa et al., 1987; Case 9, Kertesz & Benson, 1970; Ochipa & Rothi, 1989), but has also been assessed in detail in a small number of case studies. For example, Schonauer and Denes (1994) described a patient whose written spelling (in Italian) consisted mainly of pronounceable neologisms that were not obviously related to the target words, but which included a larger than random number of target letters, and which were correlated in length with the targets. Additional studies of neologistic writing (in French-speaking patients) showed that spelling was perseverative (Pillon et al., 1979) and detrimentally affected by increase in target length (Assal, 1982). Production of well-formed but neologistic kana (Japanese syllabograms) and kanji (Japanese morphograms) has also been described (Hashimoto et al., 1998; Ihori et al., 1994). Three main hypotheses have been proposed to explain jargon spelling. First, it has been attributed to the spelling system functioning in isolation, with reduced, or even without, influence of the language system. This is thought to be associated with disinhibition of writing or lowered responsethreshold (Cappa et al., 1987; Hashimoto et al., 1998; Yokoyama et al., 1981), presumably because the absence of support from the language system would otherwise preclude output. On this explanation, relevant orthographic information is not necessarily accessed. The second hypothesis is that jargon spelling arises from pathologically rapid decay of graphemic information, as in the case of graphemic buffer disorder (Hashimoto et al., 1998; Schonauer & Denes, 1994). The graphemic buffer is a temporary store where internally activated abstract graphemic representations are held while output processes (e.g., writing, typing, oral spell-

344

COGNITIVE NEUROPSYCHOLOGY , 2001, 18 (4)

ing) are executed (Caramazza, Miceli, Villa, & Romani, 1987; Ellis, 1982). On this hypothesis, correct orthographic information has been activated, but has been subject to accelerated decay. The third hypothesis is that jargon spelling is caused by weak activation of orthographic information (Ellis, Miller, & Sin, 1983; Miller & Ellis, 1987); the appropriate representations may be accessed, but the activation is insufficient to enable a correct response. This paper presents our study of a patient who produced neologistic graphemic jargon when spelling words to dictation. This jargon dysgraphia developed gradually, in the context of neurodegenerative disease, and we were able to document both the onset and the progression of the phenomenon. Through assessment of the changing relationship between target words and responses, we investigated the nature of the neologistic spelling errors.

CASE DESCRIPTION FM was a 55-year-old left-handed woman working as a care assistant in a home for the elderly, when she presented in March 1991 with a 1-year history of word-finding difficulty. According to her family, there had been no change in personality, behaviour, day -to-day memory, or practical skills. Additional details of the history have been reported elsewhere (K.S. Graham, Patterson, & Hodges, 1995; Graham et al., 1997; Hodges, Patterson, Oxbury, & Funnell, 1992; Knott, Patterson, & Hodges, 2000; Patterson & Hodges, 1992; Tyler, Moss, Patterson, & Hodges, 1997). Examination showed that FM was fully orientated, and her spontaneous speech was fluent with normal prosody, phonology, and syntax, but with prominent word-finding difficulty. Working memory, episodic memory, and visuospatial skills appeared normal, but there were mild impairments in spelling, reading, and semantic memory, as well as a severe anomia. Brain scanning at presentation with 99mTcHMPAO SPECT showed left-temporal hypoperfusion, although a computed tomography (CT) scan was normal. A T1 coronal and T2 axial


JARGON IN PROGRESSIVE DYSGRAPHIA

MRI in 1994 (3 years after presentation) revealed focal atrophy of the left temporal lobe, involving the polar and antero-lateral region, with relative sparing of the hippocampal formation and of the posterior portion of the superior temporal gyrus. At this time the right temporal pole appeared normal. Repeat MRI scans in 1996 and 1997 showed progression of the left-sided temporal lobe atrophy to involve the fusiform and posterior superior temporal gyri. In addition, the last scan showed clear-cut atrophy of the right temporal pole and a mild degree of global cerebral atrophy. As reported in our earlier paper (Graham et al., 1997), over the first 5 years of follow-up, FM’s dysgraphia, dyslexia, and anomia worsened, while other cognitive functions remained relatively stable. Speech became progressively more empty, although the patient could usually make herself understood. Practical skills and orientation continued to be unaffected, and at completion of the earlier study FM was still living alone and independently. In years 6 and 7 of our longitudinal follow-up, FM developed fairly global intellectual impairment, although with islands of preservation. Continued deterioration in her writing, reading, naming, and language comprehension led to severe communication difficulties. In autumn of 1997 (approximately 7.5 years after initial onset of symptoms), FM was moved into a residential care home where she quickly learned her way around the corridors and gardens; even at completion of this study, she still showed no evidence of spatial disorientation. She remained self-caring, and still recognised family and friends. She was able to play card games (which she had known for years) with her family, as well as simple (but newly learned) board games with staff and other residents of the home. Speech remained fluent with normal prosody, and was generally syntactically correct, but became even more empty. There was a tendency to over-use a small set of general terms such as “special,” “special place,” and “bit” (i.e., “that bit,” “those bits”), and other content words were produced only rarely. For example, her description of the Cookie Theft Picture (Goodglass & Kaplan, 1976) from the last time she was tested was as follows: “We’ve got that special place, that bit, haven’t we? And then we’ve

got that special place,” which she said as she pointed to different elements of the picture. There was no hint of jargon in her speech, and phonemic errors arose only when she attempted to repeat (spontaneously or otherwise) something that had been said to her. She remained able to repeat single monosyllabic words without difficulty, but made errors on longer words and phrases (for example, NLG: “We don’t need to do that” ® FM: “Oh, we don’t /nu/ that”).

LONGITUDINAL NEUROPSYCHOLOGICAL ASSESSMENT The longitudinal neuropsychological performance between October 1991 (6 months after presentation, when spelling was first assessed) and July 1998 is illustrated in Table 1. The test battery was given on 16 occasions at intervals of 4 to 8 months. Due to the large number of observations, and the initial stability of performance, some of the testing rounds from the first 3 years of follow-up have been omitted from the table. Because tests were not generally repeated once performance approached the floor, some were not administered in the later rounds of testing. FM’s scores are compared in Table 1 to those of 24 control subjects (approximately equivalent to FM in age and education) who were members of the MRC Cognition and Brain Sciences Unit subject panel (see Hodges & Patterson, 1995). For the Visual Object and Space Perception Battery only (VOSP, Warrington & James, 1991), performance is compared with a different set of 29 control subjects with the same characteristics and from the same source as the main control group; this occurred because the VOSP was added to our test battery subsequent to collection of the other control data. Longitudinal neuropsychological assessment (see Table 1) showed that visuoperceptual skills were relatively preserved until late in the follow-up: Copy of the Rey Complex Figure (Rey, 1941) was normal on all but the final occasion of testing, while performance on the (relatively less difficult) Screening and Dot Counting subtests of the VOSP COGNITIVE NEUROPSYCHOLOGY, 2001, 18 (4)

345

346

COGNITIVE NEUROPSYCHOLOGY , 2001, 18 (4) 121 90

Reading Surface list—regular words (126) Surface list—exception words (126) 111 77

41 48 57 3 42

69

102 79

42 48 57 2 47

63

7

46

25

-

32 8.5

Mar ’93

94 67

0 46

-

44

61

6

48

-

16 10 14

32 19

Jan ’94

71 45

39 48 59 1 44

57

6

43

19

-

31 11

Nov ’94

55 36

43 48 58 0 45

52

4

48

22

20 10 20

32 11.5

May ’95

20 12

41 48 51 0 38

49

4

47

23

-

20 10

31 10.5

Jan ’96

1/31 0/32

26 48 52 0 36

50

3

42

26

-

31 9.5

Aug ’96

-

-

b b

b

30

12 44 43

-

2

-

13

18 10 a 14

30 6

Dec’97

b

34

24 48 45

39

-

28

22

-

18 10

30 10.5

Apr ’97

b

On the final administration of the “Incomplete letters” subtest of the VOSP, written responses were accepted, as FM was no longer able to name letters. Test abandoned.

43 48 59 7 42

Semantic memory Hodges Semantic Battery Word-picture matching (48) Picture sorting—superordinate level (48) Picture sorting—subordinate level (72) Naming (48) Pyramids & Palm Trees Test (52)

a

72

6

Working memory (Wechsler Memory Scale-Revised) Forward digit span 6

Language comprehension Test for the Reception of Grammar (80)

43

39

Episodic memory Recognition Memory Test—faces (50)

25

-

29

32 12.5

Oct ’92

30 14

Oct ’91

Nonverbal problem solving Raven’s Coloured Matrices (36)

Visuoperceptual tests Rey Complex Figure (36) Copy 45 min. delayed recall VOSP Screening (20) Dot counting (10) Incomplete letters (20)

Test

Table 1. FM’s and controls’ results on general neuropsychological testing


b

b

-

11

-

2

-

12

-

19 9

24 7

Jul ’98

(1.4) (0.2) (2.2) (2.3) (1.4)

125.2 (2.7) 123.6 (3.1)

47.4 47.9 68.8 43.6 51.2

78.8 (1.8)

6.8 (1.0)

43.7 (3.8)

-

19.3 (0.9) 9.9 (0.3) 19.2 (0.8)

34.0 (3.0) 15.3 (7.4)

Controls mean (SD)




(Warrington & James, 1991) remained unaffected. Over the first 5.5 years of follow-up1 (until April 1997), FM maintained respectable performance on a test of non-verbal problem solving (Raven’s Coloured Progressive Matrices, Raven, 1962), and despite the subsequent decline, scores remained above the floor. Performance on tests of nonverbal episodic memory was inconsistent: impairment on the Recognition Memory Test for faces (Warrington, 1984) was first documented in August 1997 (nearly 6 years into follow-up), while performance on delayed recall of the Rey Complex Figure remained in the normal range for the duration. Auditory-verbal short-term memory, as measured by digit span (Wechsler, 1987), was initially normal, but had become severely impaired by completion of the study. Syntactic aspects of language comprehension (Test for the Reception of Grammar: Bishop, 1989), and semantic memory function (Semantic Battery: Hodges, Salmon, & Butters, 1992; Pyramids & Palm Trees Test: Howard & Patterson, 1992) were mildly impaired initially, and gradually deteriorated. The severe naming impairment, which was the presenting symptom, showed a relentless decline. Reading was assessed using a list of 252 words, half with regular and half with exceptional spellingsound correspondences, matched pairwise on length, initial phoneme, and word frequency (see Patterson & Hodges, 1992). FM’s reading at presentation and over the first few years of follow-up was surface dyslexic (see Table 1): She showed a significant advantage for regular over exception words, and errors were primarily phonologically plausible (e.g., pronouncing SOME to rhyme with “home”) (K.S. Graham, Hodges, & Patterson, 1994; Graham et al., 1997; Patterson & Hodges, 1992; Patterson et al., 1996). On follow-up, she began to produce an increasing number of nonphonologically plausible errors, most of which were phonologically/visually similar to the targets (e.g., SEIZE ® “siege”). In January 1996 (nearly 5 years after presentation), we observed that FM occasionally named letters in target words, rather

than attempting to read the words. This tendency increased, and by April 1997 letter naming was the only type of response which FM produced when attempting to read aloud. Her accuracy in naming these letters was initially high, but then it, too, declined. FM was ultimately completely unable to read aloud.

LETTER KNOWLEDGE AND LETTER PRODUCTION SKILLS Performance on tasks involving single letters is shown in Table 2. In our earlier report on FM (Graham et al., 1997), we documented a mild deficit in letter production, with writing in lower case being particularly affected. This deficit was demonstrated on tasks involving production of single letters without a model to copy. In each of the relevant tasks, all 26 letters in the alphabet were presented in random order, and writing in upper and lower cases was tested in blocks. On cross-case transcription, letters were presented singly, and FM was instructed to write the same letter, but in the alternate case (e.g., a ® A, G ® g). Controls scored a mean of 24.8/26 (±1.0) on UPPER ® lower, and 25.4 (±1.3) on lower ® UPPER (Graham et al., 1997). FM’s performance, which was minimally impaired on transcription into upper case and moderately impaired on transcription into lower case, fluctuated somewhat, but changed little between August 1993 and January 1998. Early in the follow-up, errors on letter transcription were mainly non-responses, but these were later outnumbered by case errors (e.g., b ® b, instead of b ® B); a small number of substitution errors (e.g., a ® N, instead of a ® A) occurred at most stages in the follow-up. Performance on writing single letters to dictation showed a similar pattern with respect to accuracy, but declined somewhat earlier in follow-up than letter transcription. The decline was associated with an increase in letter substitution errors. Letter copying remained fluent, quick, and accurate, indicating that the letter production difficulties cannot be attributed to a

1

Unless otherwise specified, the length of follow-up will be stated with reference to the first time that spelling was assessed, in October 1991 (6 months after presentation). COGNITIVE NEUROPSYCHOLOGY, 2001, 18 (4)

347


Table 2. Longitudinal performance on letter tasks


Task/Target case

Aug ’93

Letter production Letter transcription (26) ® UPPER 22 ® lower 13 Writing letters to dictation (26) UPPER 24 lower 21 Letter copying (26) UPPER 26 lower 26 Delayed letter copying (26) UPPER – lower – Letter knowledge Letter naming (26) UPPER – lower – PALPA case matching (26) lower to UPPER –

Jan ’94

May ’95

Jan ’96

Aug ’96

Apr ’97

Jan ’98

Jul ’98

21 16

23 20

21 15

23 14

18 12

19 12

a

24 19

22 16

22 15

9 7

5 6

– –

– –

26 26

26 26

26 25

26 26

26 26

26 25

26 26

– –

– –

– –

25 24

26 26

26 21

24 20

24 24

19 19

11 14

8 9

0 0

– –

– –

–

26

26

26

26

26

26

a

a

FM unable to comprehend task.

motor problem. In delayed letter copying, single letters were shown briefly, and FM was then asked to write each one without further help. Performance remained relatively preserved on this task until late in the study. The few errors involved production of a visually similar letter (e.g., q ® g, V ® U). Although FM’s ability to name single letters deteriorated relentlessly, it is noteworthy—given the severity of her anomia—that she remained able to name some letters for as long as she did. Note that letter naming and oral spelling reached the floor concurrently (in April 1997, 5.5 years into the follow-up). FM was able to recognise letters, even at completion of the study. This was demonstrated by her performance on one of the subtests in the PALPA Battery (Psycholinguistic Assessments of Language Processing in Aphasia; Kay, Lesser, & Coltheart, 1992) in which subjects are given a lower-case letter (e.g., g), and asked to indicate which of two upper-case letters (e.g., G, C), has the same identity. FM achieved perfect scores at each of the six occasions on which she did this test.

348


In summary, FM’s letter copying and letter recognition remained flawless, even at completion of this study. The initially mild impairment in letter production became more severe, and ultimately affected upper-case as well as lower-case writing. This suggests continued degradation in orthographic knowledge supporting production.

CHARACTERISATION OF LONGITUDINAL PERFORMANCE ON SPELLING-TO-DICTATION: EVOLUTION OF JARGON SPELLING Test administration FM was asked to write the same list of words to dictation at 4- to 8-month intervals (coinciding with the administration of the general neuropsychological tests), over a period of nearly 7 years. Oral spelling of the same list was assessed over a period of nearly 5 years. The target words were 18 regular (e.g., brown, shed) and 18 exception words (e.g., broad, shoe) matched pairwise on word frequency,



initial phoneme, and length (which ranged from 2 to 6 letters, with a mean of 4.5). The manipulation of regularity was pertinent to the spelling disorder (surface dysgraphia) exhibited by FM early in the course of her illness, and although this manipulation is not of particular relevance to the investigation of jargon spelling, we continued to use the same test to enable longitudinal comparison. Target words were dictated in random order and, to ensure accurate perception, we asked FM to repeat each word before responding; she had no difficulty complying with this demand, even on the final occasion of testing. Two changes in the testing procedure were instituted part-way through the study: As from January 1994 (2.25 years after the first test administration) FM was asked to repeat target words after as well as before writing (or orally spelling) them. We were motivated by the increasing number of nonphonologically plausible spelling errors to look for evidence that FM retained the target words in working memory while executing her responses. The second change was instituted as from January 1996, and entailed covering all previous responses in written spelling in order to prevent undue influence on subsequent performance; this change was motivated by the observation of perseveration in FM’s spelling.

Background FM’s performance on this test over the first 4 years of follow-up was reported in an earlier paper (Graham et al., 1997) which showed, in brief, that spelling declined, and accuracy in terms of number of words spelled completely correctly reached the floor (see Figure 1). At presentation, FM’s spelling was surface dysgraphic: She was better at spelling words that are regular, as compared to exceptional, in terms of sound-to-spelling correspondences, and her spelling errors tended to be phonologically plausible (e.g., snatch ® SNACH). Nonphonologically plausible spelling errors (referred to as “letter errors” in the earlier paper; e.g., snatch ® SNISH) gradually increased, and eventually became the predominant type of response.

Figure 1. Mean number of words spelled correctly in writing to dictation, and mean number of letters overlapping between targets and corresponding responses (regardless of serial position), for all longitudinal testing sessions.

Qualitative description Throughout the study, FM’s writing remained fluent, and seemed to be executed with normal speed, although we did not time her responses. In fact, the fluency was remarkable, given the severity of the spelling disorder. FM sometimes paused briefly, as if to think, before responding, but rarely did she hesitate in the middle of a word once she had commenced writing. All responses were printed in upper case, and non-letter characters were never produced. Occasional self-corrections were observed.

Repetition of target words after responding Until the final assessment, FM had little or no difficulty with spoken reproduction of target words after attempting to spell them: Between January 1994 (when this was instituted) and August 1996 her performance was flawless; on the next two COGNITIVE NEUROPSYCHOLOGY, 2001, 18 (4)

349



occasions, in April and December 1997, she failed to remember 7 and 3, respectively, of the 36 target items. On the final occasion of testing (July 1998), however, FM was unable to repeat any of the target items after spelling. Note that any attempt to accomplish this by means of reading the written response would have been precluded by the weak relationship between targets and responses (even had FM retained good ability to read aloud). FM’s reasonably well-preserved ability (on all but the final assessment) to repeat these words after responding demonstrates that she was generally able to maintain the target words in working memory while spelling responses were executed. Her poor spelling performance could not, therefore, be attributed to an additional problem in short-term retention. It is remarkable that FM performed as well as she did on repetition of the target words after a delay in which spelling responses were executed, particularly in oral spelling, given that testing in 1994 (K.S. Graham et al., 1995) demonstrated poor performance on repetition of single words after a 5-second delay filled with counting aloud.

Oral spelling FM was assessed longitudinally on oral spelling of the stimulus words in our main spelling test, on 10 occasions between May 1992 (7 months into follow-up) and August 1996, after which she was no longer able to perform the task. When oral spelling was last attempted (in April 1997), FM was unable to produce any letter names, and instead responded by tracing letters on the table with her finger; the test was, therefore, abandoned. Not surprisingly, the last time that FM was able to perform oral spelling coincided with the last time that she was able to name any written letters. Performance on oral spelling was qualitatively similar to, and declined in parallel with, written spelling; overall accuracy was equivalent across modalities at each assessment; all (McNemar’s) c 2 (1) values < 1.5, all p values > .1 (McNemar’s test has been used throughout for paired comparisons, while ordinary chi-square tests have been used for unpaired comparisons). Due to

350


the lack of difference between output modalities, unless explicitly noted, all further analyses have been conducted on written output.

Similarity between targets and responses— overlapping letters Inspection of the nonphonologically plausible spelling errors suggested that they became progressively less similar to the target words (e.g., scribe ® SCRIPE, January, 1992; ® SRIME, November, 1994; ® BRINCE, August, 1996; ® MEATH, November, 1997; underlining indicates letters in both target and response). To examine in detail the apparent decrease in similarity between targets and error responses, we counted the number of letters overlapping between targets and corresponding responses for each test administration. Except where noted otherwise, this measure of accuracy did not take account of letter position. Thus, scribe ® CRIST would be credited with four letters correct, even though all of the overlapping letters occupied different positions in the target and response. Figure 1, which shows the mean percentage across all test items of letters that overlapped between targets and responses, demonstrates that the onset of the jargon spelling was gradual, and that performance measured in this fashion continued to decline after overall accuracy had reached the floor. Even at the final assessment, FM managed to produce a mean of 26.1% of the letters in each target word, suggesting that her letter production could not be characterised as completely “random.” To characterise further FM’s written spelling responses, we also counted letters that appeared in targets, but not in the corresponding responses (i.e., omissions), as well as letters that appeared in responses, but not in the corresponding targets (i.e., extraneous letters). The results of these letter counts are shown in Figure 2, which contrasts them with the number of correct letters produced. As can be seen in the figure, the numbers of omitted and extraneous letters gradually increased, and ultimately outnumbered letters that overlapped with the targets.



the follow-up than suggested by the above analyses. For each time the spelling test was administered, responses were randomly assigned to target words; this procedure was repeated three times, and the number of letters overlapping between targets and “pseudoresponses” was counted each time. Paired ttests revealed a greater overlap between targets and matched responses, than between targets and randomly assigned responses, up to and including April 1997 (5.5 years into follow-up); oral and written spelling, January 1996: t(35) values > 2.1, p values < .05; oral and written spelling August 1996: t(35) values > 1.7, p £ .08; written spelling, April 1997: t(35) values > 2.1, p values < .05; written spelling, December 1997 and July 1998: t(35) val ues < 1.2, p values > .2. This analysis indicates that output continued to be constrained (albeit weakly) by the targets. Figure 2. Mean number (per response) of correct, extraneous, and omitted letters, in writing to dictation, for all longitudinal testing sessions.

Perseveration

Early in the follow-up, responses were significantly correlated in length with their respective targets. This correlation decreased over time, but remained significant (p £ .01) up to and including the 11th time FM performed the spelling test in May 1995 (3.5 years into the follow-up). The correlations (df = 34) were as follows: October 1991, .89; January 1992, .85; May 1992, .79; October 1992, .88; March 1993, .66; August 1993, .76; January 1994, .61; May 1994, .64; November 1994, .62; May 1995, .41; no subsequent correlations were significant (all p values > .1). This length effect constitutes a further demonstration of some correspondence between input and output in FM’s spelling, even when her overall accuracy was near or at floor.

The impression was that FM’s responses became more perseverative over time (despite our having instituted a policy of keeping previous responses covered from view). To measure this, for each test administration we counted the number of letters overlapping between each response and the previous response, and compared this with the overlap between each response and its target. As can be seen in Figure 3, perseveration was a relatively late feature in the progression of the spelling impairment. It increased substantially between January and August 1996 (nearly 5 years into longitudinal testing), and remained at the same high level for the duration of the follow-up. Not surprisingly, the sudden increase coincided with a jump in the overall number of extraneous letters (see Figure 2). Perseveration has also been documented in the spelling of other jargon dysgraphic patients (Pillon et al., 1979; Schonauer & Denes, 1994).

Similarity between targets and responses— greater than expected by chance?

(Occasional) violations of orthographic legality

An additional analysis (following Miller & Ellis, 1987) indicated that a significant overlap between targets and responses was retained even later into

Most of FM’s spelling responses, even in the late stages of follow-up, were pronounceable and wordlike. Examination of her responses on the main

Similarity between targets and responses— length

COGNITIVE NEUROPSYCHOLOGY, 2001, 18 (4)

351



Figure 3. Mean number of letters overlapping between responses and targets (i.e., correct letters), and between responses and previous responses (i.e., perseverations), in writing to dictation, for all longitudinal testing sessions.

spelling test in both written and oral spelling, across the period of follow-up, showed that the vast majority (all but 10, out of a total of 900) included both consonants and vowels. Moreover, 8 of the 10 responses without vowels occurred on the final administration of oral spelling (in August 1996), and may have arisen because FM’s anomia prevented her from reporting all of the letters she intended. To identify additional violations of orthographic legality, we searched all of FM’s spelling responses (in both modalities) for bigrams that do not appear in the English language. This was done irrespective of position within a word, and was determined by reference to Mayzner and Tresselt (1965), who tabled all bigrams occurring in English words of three to seven letters in length. The results demonstrated that few illegal bigrams were produced (no more than three per test administration), except in written spelling in January and August 1996. This increase in illegal letter combinations was due entirely to repeated perseveration on the bigram SR, which accounted for 5/6, and 7/8, illegal bigrams, respectively, on the two occasions. Summing across each administration of the spelling test (both written and oral), a total of 24 different

352


illegal bigrams were produced, most on one occasion only; in all, 42 responses incorporated one non-English bigram, while 4 responses incorporated two. Thus, FM used illegal bigrams in only 46/900 responses (5.1 %). A similar analysis involving trigrams (identified with reference to Mayzner, Tresselt, & Wolin, 1965; and Quinlan, 1992) showed comparable results. Inevitably, there were more illegal trigrams than bigrams, but they still occurred in only a minority of responses. Summing across each test administration (in both modalities), 93/900 responses (10.3%) incorporated illegal trigrams. Analyses of tetragrams and pentagrams did not seem warranted since many of FM’s responses comprised letter strings of these lengths and it would be uninformative to document, for example, that an incorrect five-letter response incorporates an illegal pentagram. In any case, the analyses of bigrams and trigrams provide evidence that FM’s spelling tended to follow the graphotactic rules of the language.

Letter frequency and distribution For each test administration, we counted the number of times (across all responses) that FM produced each of the 26 letters in the alphabet, and correlated this with letter frequency across the English vocabulary (taken from Baddeley, Conrad, & Thomson, 1960); the Spearman rank-order correlations (corrected for ties) were all significant, as was the correlation between the letter-distribution in the target words and letter-frequency (rho values > .6, p values < .01). This demonstrates that FM produced letters in proportion to their frequency in the language consistently throughout the followup. This observation is, of course, most informative with respect to performance late in the progression of the dysgraphia, after FM had ceased to produce correct or phonologically plausible responses. The finding of main interest is that even after spelling output ceased to be constrained by the targets, it continued to be influenced by letter frequency. A significant correlation between the distribution of letters produced by FM and that in the language was not unique to the main spelling test, and was



observed in other writing tasks (e.g., written naming and writing known and unknown words to dictation—see following). The number of different letters produced remained relatively stable over 6 years of follow-up, but declined between the penultimate (December 1997) and final (July 1998) assessments (see Figure 4). This was confirmed statistically by comparison of the number of different letters in the target words and in FM’s responses, for each occasion of testing; July 1998: c 2 (2) =13.02, p < .001; all other testing rounds, c 2 values < 1, p values > .1. We also examined the relationship between letter production and letter frequency. For this purpose, letters were deemed to be high or low frequency if they belonged to the top or bottom half, respectively, of the frequency distribution of English letters (Baddeley et al., 1960). FM consistently produced more high-frequency than lowfrequency letters (see Figure 4). In fact, on the final occasion of testing, she only produced high-frequency letters. This reduction in the number of different letters produced was also observed in other types of writing tests: at the final testing session, FM produced only 7 different letters (E, H, M, R,

Figure 4. Total number of different letters (in high– and low– frequency bands) produced in responses on writing to dictation, for all longitudinal testing sessions. Number of different letters in the target words is included for comparison purposes.

S, T, and Y) across 10 responses in self-generated writing (“Think of a word, and then write it”), all but one of which (Y) were high frequency.

Did FM retain the ability to activate any lexical information? Although we have previously attributed FM’s spelling impairment to deterioration in word-level orthographic representations (Graham et al., 1997), this does not preclude weak or partial activation of these representations. Some of FM’s error responses incorporated correctly spelled word segments that have unpredictable (or irregular) soundto-spelling correspondences, suggesting that at least some word-specific knowledge had been activated (Ellis, 1982, called these errors of “partial lexical knowledge”). To measure this, for each test administration we counted the phonemes with unpredictable spellings that FM had spelled correctly in her error responses. For this purpose, a phoneme was deemed to have an unpredictable spelling if it met the following criteria: (1) across the language, the phoneme is spelled in that way at that position within a syllable (i.e., initial, medial, final) no more than 20.0% of the time; and (2) the phoneme is not spelled in the most common manner for that position within a syllable (this was determined with reference to Hanna, Hanna, Hodges, & Rudorf, 1966). Results showed that until August 1996 (nearly 5 years into the followup) a mean of 200% of FM’s error responses (in both modalities) incorporated at least one correctly spelled phoneme with an unpredictable spelling. There was no evidence that word-specific knowledge was accessed subsequent to this time. The last time that FM managed to spell any word completely correctly was in May 1995 (when she scored 1, and 4/36, on written and oral spelling, respectively), while the last time that she spelled any unpredictable phonemes correctly was in August 1996, 15 months later. This suggests that FM was sometimes able to access partial word-specific knowledge in spelling, even after her overall accuracy had reached the floor. COGNITIVE NEUROPSYCHOLOGY, 2001, 18 (4)

353



Incomplete activation vs. decay It is possible that when FM first heard a spoken word in the spelling-to-dictation task, its correct orthographic representation was fully activated, but this activation decayed too quickly to enable an accurate response. Pathologically rapid decay of information should exert a progressively greater detrimental effect on performance from the beginning to the end of a response, since letters near the end of a word will be subject to greater decay than earlier ones (Katz, 1991; Miceli, Silveri, & Caramazza, 1987; Miller & Ellis, 1987). We investigated this possibility by looking for a serial position effect in FM’s spelling. Comparison across words of varying lengths was facilitated by “normalising” the length of stimuli and responses to five letters, in accordance with a widely used procedure developed by Wing and Baddeley (1980). In this procedure, letters are assigned to one of five serial positions, such that the distribution around the centre position will be symmetric (e.g., r--u--g, a-p-pe-a-r, br-a-c-k-et). The following examples demonstrate both how letters were apportioned between the five positions, and how overlap between targets and responses was determined: “t-o-a-s-t” ® T-R-OW -S-T is correct at positions 1, 4, and 5, while “d-o--w-n” -> S-O--A-M is correct at position 2 only, and “c-o-u-c-h” ® CH-OR-S-HS is correct at position 2, and partially correct at positions 1 and 5 (a score of 0.5 was assigned for partial matches). We computed the mean number (across error responses) of letters correct at each serial position, for each time FM was assessed on our spelling test, and compared this to the number (at each serial position) that would be expected by chance. The chance scores for each letter position were calculated separately for each test administration, in the following manner: (total number of letters produced correctly/total number of target letters) × (number of target letters in the relevant position). The observed and chance distributions were compared using chi-square tests, and were found to differ significantly for oral spelling on four occasions (November 1994 through August 1996); all c 2 (4) values > 9, p values < .05, and for written spelling on

354


two occasions; November 1994: c 2(4) = 11.2, p < .05; January 1996: c 2(4) = 10.3, p < .05; no trend towards a significant serial position effect was shown on any of the other occasions; all c 2 (4) values < 8, p values > .1. Because the greatest difference between observed and expected frequencies occurred at position 1, we repeated the analyses with this position excluded; this time there was no difference between the observed and chance distributions; all c 2 (4) values < 6, p values > .1. Taken together, these results indicate that on a minority of occasions FM performed better on initial letters than at subsequent positions, but performance did not decline across positions 2 through 5, suggesting that decay is probably not an adequate explanation of FM’s deficit. Note also that this pattern of results is inconsistent with that expected from damage to the graphemic buffer, in which errors are expected to occur more often in the middle of a word than at either end (Caramazza et al., 1987).

(Lack of) anosognosia It has been argued that anosognosia is a reliable, or even necessary feature of jargon aphasia (see Ryalls, Valdois, & Lecours, 1988), and it has also been reported in patients with jargon spelling (Assal, 1982; Cappa et al., 1987; Ihori et al., 1994; Pillon et al., 1979). FM, however, seemed to retain insight into her dysgraphic impairment, suggesting that anosognosia is not a necessary feature of jargon spelling. Insight was demonstrated on the writtennaming task (see below), and also by FM’s comments during spelling to dictation. For example, in August 1993 (nearly 3 years into the follow-up), she noted “I know that isn’t right, but it’s the best I can do,” while in August 1996, she commented (in her by then more anomic fashion) “I know I don’t do the things properly.” We also note that, while hypergraphia (excessive writing) is sometimes associated with jargon writing (Basso et al., 1985; Ihori et al., 1994; Lecours & Rouillon, 1976), FM showed no sign of this phenomenon.


FURTHER INVESTIGATIONS OF SPELLING


Written naming vs. writing to dictation Although FM produced jargon in spelling, there was no hint of it in her oral naming or spontaneous speech. In oral naming, she was usually unable to respond, and informal observation suggested that she was no more likely to produce a response in written than in oral naming. Her performance on the tasks of spelling and naming differed, however, on a potentially relevant characteristic: In spelling FM always produced a response, whereas in naming (both oral and written) she responded only when she thought she knew the answer. In order to make the tasks more comparable, we attempted to manipulate whether FM knew the correct answer when she responded. This resulted in four test conditions (see Table 3): (1) writing to dictation, administered in the standard way (described above); (2) writing to dictation, with the instruction to respond only if the response would be correct; (3) written naming, administered in the standard way (the patient was shown a line-drawing of an object, and was asked to write the name of the object); and (4) written naming, with coaxing (on each item) to produce a response. The identical set of 60 words/ pictures (taken from Snodgrass & Vanderwart, 1980) was used in the four test conditions, which were administered on separate days in the order in which they are listed above. Because we wanted to maximise the possibility that FM would be able to

produce a response in written naming (recall that she was severely anomic), we included only those items which she had, in earlier testing rounds, named orally at least twice. This experiment was performed on two occasions, first in August 1996 and again in April 1997 (at approximately 5, and 5.5 years into follow-up, respectively). Consistent with the findings on the main spelling test, overall accuracy in each test condition was very poor: In August 1996, FM spelled two words correctly in writing to dictation with standard administration, and only one in each of the other three test conditions, while in April 1997, she spelled no words correctly in any test condition. The measure of main interest, however, was the number of responses produced in each condition (see Table 3). FM produced many fewer responses in writing to dictation when asked to respond only if she thought she could produce the correct answer, than when no instruction in this regard was given. This demonstrates both that FM normally generated responses on the basis of incomplete information, and that she retained insight into her spelling impairment. With written naming, on the first occasion of testing the number of responses produced did not increase when FM was coaxed, and we could not induce her to respond unless she thought she knew the answer. This indicates that she needed either to know the target word, or to be given the phonology (as in dictation), in order to produce a written response. The second time that FM performed

Table 3. Test conditions and results from comparison of writing to dictation and written naming

Test conditions

Results: August 1996

Results: April 1997

a

a

Task

Respond ————————————————————————————————– On every item Only if answer is “known”

Writing to dictation

Standard administration

Written naming

Coaxed

Respond only if she thinks she can produce correct answer Standard administration

Writing to dictation Written naming

60 4

23 5

Writing to dictation Written naming

60 53

4 3

Results reported are number of responses attempted in each test condition (max. = 60).

a


355



written naming (8 months later), however, coaxing did induce her to respond to most of the test items, indicating that she was willing/able to generate a written response without phonological information. (Because of her severe anomia, we presume that FM was unable to access internally the phonology of the target words.) In summary, as FM’s disease progressed, her pattern of performance in written naming changed from one of virtually total dysgraphia to jargon dysgraphia.

The role of semantics in FM’s spelling By analogy with the way in which semantic information is hypothesised to influence processing in reading aloud (Hillis & Caramazza, 1991, 1995; Patterson & Behrmann, 1997), activation of orthographic representations in spelling to dictation, particularly for words that are processed less efficiently (either because they are infrequent or incorporate atypical sound-to-spelling correspondences), might benefit from interaction with semantic representations (Graham, Patterson, & Hodges, 2000). We looked for evidence of semantic influence in FM’s spelling by assessing her performance on “known” and “unknown” words (following a technique previously used in assessment of spoken word production, and of reading aloud, e.g., Funnell, 1996; K.S. Graham et al., 1994; Hillis & Caramazza, 1991, 1995; Knott, Patterson, & Hodges, 1997; Patterson, Graham, & Hodges, 1994; Warrington, 1975). Words classified as “known” were content words that had recently been used appropriately in FM’s spontaneous speech; these were matched pairwise with “unknown” words on word frequency (Kuçera & Francis, 1967), part of speech, and length (number of letters and number of syllables). By the time that this test was performed (approximately 4 years into the follow-up), FM’s semantic memory impairment was already rather severe, and we did not, therefore, seek independent evidence that she no longer comprehended the “unknown” words, which were selected purely on the basis that they matched the “known” words, on the criteria listed earlier. The test comprised a total of 90 words (45 “known”, e.g., suddenly, church, decent; and 45

356


“unknown”, e.g., properly, school, savage), which were dictated in random order. As with the other spelling tasks, FM’s overall accuracy was near the floor (she correctly spelled only three words, all of which were “known”: tea, love, church), and errors were all nonphonologically plausible, but showed some overlap with the targets (e.g., birthday ® BRISH, quick ® CRIM). Because performance was so impaired, the analyses focused on the number of letters that overlapped between targets and responses. A paired t-test showed a trend towards better performance on “known” than “unknown” words; mean number of letters overlapping: known 2.2, unknown 1.7; t(44) = 1.9, p = .06. This suggests that knowing a word gave a small boost to FM’s attempt to write it, and implies that the interaction between meaning and orthography may have provided some support for spelling. It is clear, however, that this was not sufficient to enable fully accurate spelling, and that its beneficial influence on partially correct output was limited. The finding that FM was even marginally better at spelling “known” than matched “unknown” words would be unexpected if FM’s spelling deficit arose mainly from a peripheral locus. Instead, it suggests that she had a central spelling deficit.

DISCUSSION This paper reports our longitudinal study of FM, a patient with acquired, and progressive, dysgraphia. The study continues an earlier investigation (Graham et al., 1997) which documented a 4-year progression from initial surface dysgraphia, in which errors were primarily phonologically plausible, through an increasing tendency to produce nonphonologically plausible spelling errors which, after 4 years of follow-up, dominated spelling performance. At commencement of the present investigation FM was virtually unable to spell any words correctly, or even in a phonologically plausible manner. Nevertheless, we were able to document continuing deterioration, as evidenced by the reduction in letters overlapping between targets



and responses, over the next 2.5 years. At completion of this study, written output was word-like and fluently executed, but grossly inaccurate. Furthermore, the apparently random strings of letters produced in writing to dictation were all well formed, and free of non-letter characters. This fluent production of written neologisms that followed the graphotactic rules of the language led us to label this pattern of performance “jargon spelling.” Because the jargon spelling occurred in the context of progressive disease, we were able to study it at different stages, and to look retrospectively at the onset. In FM’s spelling to dictation, the number of correct letters (i.e., those which overlapped between targets and their corresponding responses) showed a gradual decline, dropping from 95.7% correct to 26.1% over approximately 6.5 years. Eventually, responses incorporated more extraneous than correct letters, although a greater than random correspondence between targets and responses was maintained until late in the follow-up. Perseveration was a late feature, though it dominated FM’s performance once it emerged (see Figure 3). Letters were consistently produced in proportion with their frequency in the language, but the number of different letters produced ultimately declined, and was finally restricted to a small set of (mainly) high-frequency letters. The rarity of violations of the graphotactic rules of the language, in conjunction with the significant effect of letter frequency, demonstrates that (contrary to appearances) FM’s letter production even at completion of the study was not random, and was influenced by premorbid spelling knowledge, though no longer word-specific. Jargon spelling could arise following normal activation of orthographic representations if the activation then decayed too rapidly to enable accurate responses. The finding that FM’s performance tended to be better on initial letters than at other serial positions within the word (although only on a minority of occasions) is consistent with this hypothesis, but other aspects of performance are not, including the lack of change in performance across serial positions from 2–N, and the known/ unknown word effect. In addition, to make this hypothesis viable, one would have to explain why

FM was unable to reactivate (subsequent to decay) the necessary orthographic information to complete a partially executed response—particularly in written spelling, where the partially completed response remains available for inspection. Our interpretation of FM’s spelling impairment, although we acknowledge it to be a rather underspecified one, is a gradually worsening deficit in activating orthographic representations. Initially FM showed surface dysgraphia, and in some views of the lexical system (e.g., the triangle model, developed by Seidenberg & McClelland, 1989) words with unpredictable sound-to-spelling correspondences are expected to be most vulnerable to reduction in activation. The subsequent nonphonologically plausible errors could arise from more severe disruption. A similar pattern of performance was observed following damage to two implemented models of spelling in which familiar and novel words are spelled via one procedure. In Olson and Caramazza’s (1994) network mild damage led to phonologically plausible (or in their terminology, “regularisation”) errors, while more severe damage led to more errors, most of which were nonphonologically plausible. Bullinaria (1997) also implemented a model of spelling in which mild damage led to regularisation errors on exception words, while processing of regular words was not affected until damage was more severe. As noted in our earlier paper (Graham et al., 1997), this parallels what happened to FM’s spelling over the first 3 to 4 years of follow-up. As FM’s spelling continued to deteriorate, the declining overlap between targets and responses suggested increasingly weak activation of relevant orthographic information. Even after spelling had become severely impaired, however, there was evidence that output continued to be at least weakly constrained by semantics and/or phonology. First of all, at approximately 4 years into the follow-up the overlap between targets and responses was greater on “known” than “unknown” words, where the “known” words are ones that FM could still produce appropriately in spontaneous speech, which indicates successful activation of both semantic and phonological representations for these items. Furthermore, at 5 years into the study, when FM was COGNITIVE NEUROPSYCHOLOGY, 2001, 18 (4)

357



unable to respond at all in a written picture-naming task, she was still able to respond to the same test items when she was given the phonology (i.e., the word was dictated to her). Although the resultant responses displayed only partial overlap with the target words, the fact that she was able to respond suggests that phonological activation supported orthographic output. Further evidence for phonological influence upon spelling came from the main writing to dictation test: a greater-than-random overlap between targets and responses was maintained until 5.5 years into the follow-up. Thus, hearing (and repeating) spoken target words continued to influence orthographic output. Eventually, even the well-learned correspondences between phonology and orthography ceased to control FM’s spelling behaviour, and on the final two occasions of testing (approximately 6 years after the initial assessment) there was no detectable relationship between targets and responses in writing to dictation. Even at this stage, however, written output could not be characterised as random strings of letters: Performance continued to be influenced by letter frequency and the graphotactic rules of the language, and had come (particularly over the final 2 years of the study) to be influenced by recency of activation, as demonstrated by perseverative responding. This suggests that performance was governed more by “background” activation in orthography than by input from phonology or semantics. The letter-frequency effect could arise because more frequent letters have lower activation thresholds. The finding that few illegal letter combinations were produced demonstrates that context remained a significant constraint in FM’s performance even after spelling had become severely impaired, and highlights its importance in modelling of spelling (for discussion see Bullinaria, 1997; Olson & Caramazza, 1994). The observation of perseveration suggests that recently produced letters retained relatively higher activation, leading them to be incorporated into subsequent responses. It has previously been suggested that jargon dysgraphia can be explained by the disinhibition of writing from the influence of language functions (Cappa et al., 1987; Hashimoto et al., 1998;

358


Yokoyama et al., 1981), but this hypothesis does not seem apt in FM’s case. The idea of disinhibition arising at a motoric level, (i.e., that motor engrams for writing are no longer controlled by the language areas of the brain) (Yokoyama et al., 1981), is incompatible with the findings, since this could not explain the observation of jargon dysgraphia in oral, as well as written, spelling. Furthermore, the jargon spelling could not be attributed to general disinhibition, since there was no jargon in FM’s speech. It has also been argued that a lowered response threshold is necessary to explain why a subject generates responses on the basis of incomplete information (Cappa et al., 1987; Hashimoto et al., 1998). Although this seems tautological (i.e., the writing of dysgraphic subjects is based on incomplete information because they are dysgraphic), and therefore not explanatory, we note that there was evidence that FM’s response threshold did change: at 5 years into the follow-up, she produced no responses in a written naming task, whereas 6 months later, she willingly produced (erroneous) responses. An important, and as far as we are aware novel, feature of our data concerning FM’s jargon dysgraphia is that her willingness to produce written neologisms was manipulable by instruction. In writing to dictation, where FM’s natural (uninstructed) tendency was to produce a response on every trial, the instruction to “respond only if you think that you can produce a correct response” resulted in a shift to a significant proportion of trials with no responses. FM’s behaviour in written naming was also susceptible to instruction, but in the opposite direction. Initially FM produced few responses in written naming, even when coaxed to respond. Six months later, her natural (uninstructed) tendency was still to produce nothing on almost every trial of written naming; but at this stage, coaxing her to respond led her to write neologistic jargon for the majority of items. The modulation of FM’s response rate by instruction points to the general importance of response criterion. Across different patients, or even within one patient across different tasks, it seems clear that a patient’s willingness to respond—even when the response may be completely neologistic and in fact



bear virtually no relation to the target—is not a fixed property of the damaged language system. To the best of our knowledge, this is the first report documenting jargon spelling in a patient with progressive disease. This provided a unique opportunity to document the onset, and to study the deficit at different stages. Also as far as we know, and rather surprisingly, this is the first detailed case study of this syndrome in an Englishspeaking patient. The overall picture appears to be one of gradual loss of semantic and phonological influence on spelling output, while the most basic, general features of orthographic knowledge— reflected in letter frequencies and co-occurring letter combinations—were retained. Manuscript received 20 January 1999 Revised manuscript received 30 June 2000 Revised manuscript accepted 17 July 2000

REFERENCES Alexander, M.P., Fischette, M.R., & Fischer, R.S. (1989). Crossed aphasias can be mirror image or anomalous. Brain, 112, 953–973. Assal, G. (1982). Étude neuropsychologique d’une aphasie croisée avec jargonagraphie. Revue Neurologique, 138, 507–515. Baddeley, A.D., Conrad, R., & Thomson, W.E. (1960). Letter structure of the English language. Nature, 186, 414–416. Basso, A., Capitani, E., Laiacona, M., & Zanobio, M.E. (1985). Crossed aphasia: One or more syndromes? Cortex, 21, 25–45. Basso, A., Taborelli, A., & Vignolo, L.A. (1978). Dissociated disorders of speaking and writing in aphasia. Journal of Neurology, Neurosurgery, and Psychiatry, 41, 556–563. Bishop, D.V.M. (1989). Test for the Reception of Grammar (2nd ed.). Cambridge, UK: Medical Research Council. Bramwell, B. (1899). On “crossed” aphasia and the factors which go to determine whether the “leading” or “driving” speech-centres shall be located in the left or in the right hemisphere of the brain, with notes of a

case of “crossed” aphasia (aphasia with right-sided hemiplegia) in a left-handed man. The Lancet, 1, 1473–1479. Bullinaria, J.A. (1997). Modeling reading, spelling, and past tense learning with artificial neural networks. Brain and Language, 59, 236–266. Cappa, S.F., Cavallotti, G., & Vignolo, L.A. (1987). Jargonagraphia: Clinical and neuropsychological correlates. Neuropsychologia, 25, 281–286. Caramazza, A., Miceli, G., Villa, G., & Romani, C. (1987). The role of the graphemic buffer in spelling: Evidence from a case of acquired dysgraphia. Cognition, 26, 59–85. Demeurisse, G., Hublet, C., Coekaerts, M.J., Derouck, M., & Capon, A. (1986). Assessment of hemispheric dominance for language in crossed aphasia by two atraumatic methods. Cortex, 22, 305–311. Ellis, A.W. (1982). Spelling and writing (and reading and speaking). In A.W. Ellis (Ed.), Normality and pathology in cognitive functions (pp. 113–146). London: Academic Press. Ellis, A.W., Miller, D., & Sin, G. (1983). Wernicke’s aphasia and normal language processing: A case study in cognitive neuropsychology. Cognition, 15, 111– 144. Funnell, E. (1996). Response biases in oral reading: An account of the co-occurrence of surface dyslexia and semantic dementia. Quarterly Journal of Experimental Psychology, 49A, 417–446. Goodglass, H., & Kaplan, E. (1976). The assessment of aphasia and related disorders. Philadelphia, PA: Lea & Febiger. Graham, K.S., Hodges, J.R., & Patterson, K. (1994). The relationship between comprehension and oral reading in progressive fluent aphasia. Neuropsychologia, 32, 299–316. Graham, K.S., Patterson, K., & Hodges, J.R. (1995). Progressive pure anomia: Insufficient activation of phonology by meaning. Neurocase, 1, 25–38. Graham, N.L., Patterson, K., & Hodges, J.R. (1997). Progressive dysgraphia: Co-occurrence of central and peripheral impairments. Cognitive Neuropsychology, 14, 975–1005. Graham, N.L., Patterson, K., & Hodges, J.R. (2000). The impact of semantic memory impairment on spelling: Evidence from semantic dementia. Neuropsychologia, 38, 143–163. Habib, M., Joanette, Y., Ali-Cherif, A., & Poncet, M. (1983). Crossed aphasia in dextrals: A case report with special reference to site of lesion. Neuropsychologia, 21, 413–418. COGNITIVE NEUROPSYCHOLOGY, 2001, 18 (4)

359



Hanna, R.R., Hanna, J.S., Hodges, R.E., & Rudorf, E.H. (1966). Phoneme-grapheme correspondences as cues to spelling improvement. Washington, DC: US Government Printing Office. Hashimoto, R., Tanaka, Y., & Yoshida, M. (1998). Selective kana jargonagraphia following right hemispheric infarction. Brain and Language, 63, 50– 63. Hillis, A.E., & Caramazza, A. (1991). Mechanisms for accessing lexical representations for output: Evidence for a category specific semantic deficit. Brain and Language, 40, 106–144. Hillis, A.E., & Caramazza, A. (1995). Converging evidence for the interaction of semantic and sublexical phonological information in accessing lexical representations for spoken output. Cognitive Neuropsychology, 12, 187–227. Hodges, J.R., & Patterson, K. (1995). Is semantic memory consistently impaired early in the course of Alzheimer’s disease? Neuroanatomical and diagnostic implications. Neuropsychologia, 33, 441–459. Hodges, J.R., Patterson, K., Oxbury, S., & Funnell, E. (1992). Semantic dementia: Progressive fluent aphasia with temporal lobe atrophy. Brain, 115, 1783– 1806. Hodges, J.R., Salmon, D.P., & Butters, N. (1992). Semantic memory impairment in Alzheimer’s disease: Failure of access or degraded knowledge? Neuropsychologia, 30, 310–314. Howard, D., & Patterson, K. (1992). Pyramids and Palm Trees: A test of semantic access from pictures and words. Bury St Edmunds, UK: Thames Valley Test Company. Ihori, N., Kashiwagi, T., Kashiwagi, A., & Tanabe, H. (1994). Jargonagraphia in kanji and kana in a Japanese crossed Wernicke’s aphasic. Brain and Language, 47, 197–213. Katz, R.B. (1991). Limited retention of information in the graphemic buffer. Cortex, 27, 111–119. Kay, J., Lesser, R., & Coltheart, M. (1992). Psycholinguistic Assessments of Language Processing in Aphasia. Hove, UK: Lawrence Erlbaum Associates Ltd. Kertesz, A., & Benson, D.F. (1970). Neologistic jargon: A clinicopathological study. Cortex, 6, 362–386. Kinsbourne, M., & Warrington, E.K. (1963). Jargon aphasia. Neuropsychologia, 1, 27–37. Knott, R., Patterson, K., & Hodges, J.R. (1997). Lexical and semantic binding effects in short-term memory: Evidence from semantic dementia. Cognitive Neuropsychology, 14, 1165–1216.

360


Knott, R., Patterson, K., & Hodges, J.R. (2000). The role of speech production in auditory–verbal shortterm memory: Evidence from progressive fluent aphasia. Neuropsychologia, 38, 125–142. Kuçera, H., & Francis, W.N. (1967). Computational analysis of present-day American English. Providence, RI: Brown University Press. Lecours, A.R., & Rouillon, F. (1976). Neurolinguistic analysis of jargonaphasia and jargonagraphia. In H. Whitaker & H.A. Whitaker (Eds.), Studies in neurolinguistics, Vol. 2 (pp. 94–114). London: Academic Press. Mayzner, M.S., & Tresselt, M.E. (1965). Tables of single-letter and digram frequency counts for various word-length and letter-position combinations. Psychonomic Monograph Supplements, 1, 13–32. Mayzner, M.S., Tresselt, M.E., & Wolin, B.R. (1965). Tables of trigram frequency counts for various wordlength and letter-position combinations. Psychonomic Monograph Supplements, 1, 33–78. Miceli, G., Silveri, M.C., & Caramazza, A. (1987). The role of the phoneme-to-grapheme conversion system and of the graphemic output buffer in writing. In M. Coltheart, G. Sartori, & R. Job (Eds.), The cognitive neuropsychology of language (pp. 235–252). Hove, UK: Lawrence Erlbaum Associates Ltd. Miller, D., & Ellis, A.W. (1987). Speech and writing errors in “neologistic jargonaphasia”: A lexical activation hypothesis. In M. Coltheart, G. Sartori, & R. Job (Eds.), The cognitive neuropsychology of language (pp. 253–271). Hove, UK: Lawrence Erlbaum Associates Ltd. Ochipa, C., & Rothi, L.J.G. (1989). Recovery and evolution of a subtype of crossed aphasia. Aphasiology, 3, 465–72. Olson, A., & Caramazza, A. (1994). Representation and connectionist models: The NETspell experience. In G.D.A. Brown & N.C. Ellis (Eds.), Handbook of spelling: Theory, process and intervention (pp. 337– 363). Chichester: John Wiley & Sons Ltd. Patterson, K., & Behrmann, M. (1997). Frequency and consistency effects in a pure surface dyslexic patient. Journal of Experimental Psychology: Human Perception and Performance, 23, 1217–1231. Patterson, K., Graham, N., & Hodges, J.R. (1994). The impact of semantic memory loss on phonological representations. Journal of Cognitive Neuroscience, 6, 57– 69. Patterson, K., & Hodges, J.R. (1992). Deterioration of word meaning: Implications for reading. Neuropsychologia, 30, 1025–1040.



Patterson, K., Plaut, D.C., McClelland, J.L., Seidenberg, M.S., Behrmann, M., & Hodges, J.R. (1996). Connections and disconnections: A connectionist account of surface dyslexia. In J. Reggia, R. Berndt, & E. Ruppin (Eds.), Neural modelling of cognitive and brain disorders (pp. 177–199). Singapore: World Scientific. Pillon, B., Desi, M., & Lhermitte, F. (1979). Deux cas d’aphasie croisée avec jargonagraphie chez des droitiers. Revue Neurologique, 135, 15–30. Quinlan, P.T. (1992). Oxford psycholinguistic database. Oxford: Oxford University Press. Raven, J.C. (1962). Coloured Progressive Matrices Sets A, AB, B. London: H.K. Lewis. Rey, A. (1941). L’examen psychologique dans les cas Archives de d’encephalopathie traumatique. Psychologie, 28, 286–340. Ryalls, J., Valdois, S., & Lecours, A.R. (1988). Paraphasia and jargon. In F. Boller & J. Grafman (Eds.), Handbook of neuropsychology, Vol. 1, (pp. 367– 376). Amsterdam: Elsevier Science Publishers. Schonauer, K., & Denes, G. (1994). Graphemic jargon: A case report. Brain and Language, 47, 279–299. Seidenberg, M.S., & McClelland, J.L. (1989). A distributed, developmental model of word recognition and naming. Psychological Review, 96, 523–568. Snodgrass, J.G., & Vanderwart, M. (1980). A standardized set of 260 pictures: Norms for name agreement, image agreement, familiarity and visual complexity.

Journal of Experimental Psychology: Human Learning and Memory, 6, 174–215. Tyler, L.K., Moss, H.E., Patterson, K., & Hodges, J. (1997). The gradual deterioration of syntax and semantics in a patient with progressive aphasia. Brain and Language, 56, 426–476. Warrington, E.K. (1975). Selective impairment of semantic memory. Quarterly Journal of Experimental Psychology, 27, 635–657. Warrington, E.K. (1984). Recognition Memory Test. Windsor, UK: NFER Nelson. Warrington, E.K., & James, M. (1991). The Visual Object and Space Perception Battery. Bury St. Edmunds, UK: Thames Valley Test Company. Washimi, Y., Makishita, H., Miyasaka, M., Fujita, T., & Yanagisawa, N. (1987). Crossed aphasia with jargonagraphia due to right putaminal hemorrhage— 133 A study by Xe single photon emission CT (English language abstract to Japanese article). Clinical Neurology (Tokyo), 27, 12–18. Wechsler, D.A. (1987). Wechsler Memory Scale—Revised. San Antonio, TX: Psychological Corporation. Wing, A.M., & Baddeley, A.D. (1980). Spelling errors in handwriting: A corpus and a distributional analysis. In U. Frith (Ed.), Cognitive processes in spelling (pp. 251–285). London: Academic Press. Yokoyama, K., Okubo, M., Doseki, K., & Yamadori, A. (1981). Jargonagraphia in crossed motor aphasia (English language abstract to Japanese article). Clinical Neurology (Tokyo), 21, 961–967.


361