Neglect and Visual Language
Department of Psychology, University of York, York, UK
Department of Psychology, University of Durham, Durham, UK
Department of Psychology, University of Lancaster, Lancaster, UK
Introduction
Reading and writing are among the functions commonly affected in unilateral neglect. A patient with left-side neglect following a right parietal lesion will tend to omit the initial words of lines of text, and will make errors on single words which affect the initial letters, for example misreading WINE as “mine” or MESSAGE as “passage”. With the exception of Kinsbourne and Warrington (1962), the only treatments of the impact of neglect on reading one can find before the mid-1980s are embedded in more general accounts of the “neglect syndrome” (e.g. Assal & Zander, 1969; Diller & Weinberg, 1977; Friedland & Weinstein, 1977; Gilliat & Pratt, 1952).
The lessons to have been learnt from examining neglect dyslexia through the cognitive neuropsychologist’s microscope are, in fact, the lessons that nearly always seem to follow from close inspection of traditional neuropsychological syndromes. The first is that the condition comes in different forms with different properties; the second that at least some of these different forms can dissociate from other symptoms (in this case, other manifestations of unilateral neglect) with which they are commonly associated. Our interpretation and understanding of unilateral neglect, including the effect of neglect on reading and writing, is undergoing major re-assessment and re-evaluation. It follows that any conclusions reached must be regarded as tentative and liable to modification in the light of further investigations.
Neglect and Reading I: Spatial Neglect Dyslexia
Spatial Neglect Dyslexia: A Basic Description
Following a series of cerebrovascular accidents, patient V.B. (Ellis, Flude, & Young, 1987a) showed left-side neglect on a wide range of tasks including drawing, line bisection and cancelling letters in arrays. Although her speech production and comprehension were unimpaired, her reading and writing were affected. When reading passages of text, she often failed to attempt words at the beginnings of lines and made errors on a proportion of the words she did attempt.
In their analysis of V.B.’s neglect dyslexia, Ellis et al. (1987a) concentrated on the errors she made in reading single words. From this, it would be easy to convey the impression that V.B. misread most of the words she attempted. She did not. Given unlimited time, she read over 90% of single words correctly. In a number of tests, Ellis et al. were unable to find evidence that V.B.’s accuracy was affected by the frequency of occurrence of stimulus words, their imageability, grammatical class or spelling-sound regularity. Errors showed no significant tendency to be of higher frequency than the target words that elicited them.
The majority of V.B.’s errors involved the initial (left-most) letters of words. Sometimes this involved the simple deletion of initial letters from the target. Examples (in which the target word is shown in capital letters to the left of the arrow and V.B.’s error is to the right) include CLOVE 
love, DEARTH earth, CHAMBER amber and DRAUGHT aught. Other errors involved the addition of letters to the beginning of the target word (e.g. ATE date, LASS glass). More commonly, however, V.B. substituted letters of the target word with other letters. In the majority of cases, this substitution involved the initial letter only (e.g. EAR car, SOON moon, HADDOCK paddock), but sometimes two or three initial letters were involved (e.g. CABIN robin, YELLOW pillow).
Because the prototypical error for V.B. involved substituting letters from the target word with the same number of letters in the error, her misreadings tended to contain the same number of letters as the target words. Examples involving quite different numbers of letters did occur from time to time (e.g. SOCK hassock, HEIRLOOM bloom), but overall there was a clear correlation between target length and error length, a phenomenon which had been commented upon previously by Kinsbourne and Warrington (1962).
Ellis et al. (1987a) adopted a fairly strict criterion for what should count as a neglect error for purposes of analysis. This criterion was that the target and error should be identical to the right of an identifiable “neglect point” in each word and should have no letters in common to the left of those two points. Thus in FABLE table, the neglect point comes after the F in FABLE and after the t in table, whereas in YELLOW pillow it comes after the E in YELLOW and the i in pillow. Altogether, 66% of V.B.’s errors satisfied this strict criterion, 26% were labelled “visual” (e.g. TSAR star, MARQUEE quarrel) and the remaining 8% involved the production of nonwords as responses (e.g. LATCH flatch, SLAVE blave).
If V.B. was asked to read a word and also explain what it meant, the definitions she provided were always appropriate to any misreading she made. Thus, she misread RICE as “price” and defined it as “how much for a paper or something in a shop”, misread LIQUID as “squid” and defined it as “a kind of sea creature”, and misread CLOVER as “lover”, producing the definition “partner, or someone you have an affair with; a sweetheart”.
V.B.’s identification of words spelled aloud to her was excellent. Her accuracy and error pattern could be changed dramatically by presenting written words in formats other than conventional, horizontal, left-to-right. If a passage of text was rotated clockwise through 90° so that the lines ran from top to bottom rather than from left to right, her performance improved considerably. In particular, she no longer omitted the initial words on each line. If single words were presented upside-down (i.e. rotated through 180°), she managed to read over 65% correctly. Importantly, her neglect errors were now concentrated on the final letters, which are on the left when words are upside down, (e.g. PLANT plane, SCORN score). Thus, what mattered for V.B.’s neglect dyslexia was the spatial (left-to-right) location of letters in words rather than their first-to-last position. Hence we shall refer to this variety of neglect dyslexia as spatial neglect dyslexia.
Length and Lexicality
Behrmann, Moscovitch, Black and Mozer (1990) studied two patients with neglect dyslexia. One of their patients, A.H., would appear to have displayed much the same form of neglect dyslexia as V.B. A.H. made neglect dyslexic errors in the left-most letters of words presented either normally or mirror-reversed. Errors to words displayed vertically were not concentrated on the initial letters. Forty-two per cent of words in conventional format were read correctly, and 61 % of errors were neglect errors.
Behrmann et al. (1990) asked A.H. to read words that were either normally spaced or extended by the addition of an extra character space between the letters (cf. HOUSE and H O U S E). Reading accuracy was significantly worse for the extended words than for the normal words. In a study not reported in Ellis et al. (1987a), V.B. was also asked to read words which were either normal or extended. She read 30/40 normal words correctly and 25/40 extended words. While not conclusive, this result is at least compatible with Behrmann and co-workers’ finding, and supports the notion that in spatial neglect dyslexia, the initial letters of words are more likely to suffer the effects of neglect (and hence words are more likely to be misread) if the words occupy a greater horizontal extent.
If that is the case, then long words may tend to be read less well than short words. Behrmann et al. (1990) showed that A.H.’s reading accuracy declined from 80% correct to 67% correct across a range of lengths from three to nine letters. Ellis et al. (1987a) failed to obtain a reliable length effect with V.B., but used a narrower range of lengths. Their stimuli were also handwritten, and there may have been a tendency to space out the letters of short words more than long words. There is another factor which may militate against obtaining a length effect. Ellis et al. (1987a) showed that V.B.’s typical neglect error involved the first one or two letters of words, irrespective of their length. Longer words will typically be easier to guess from all except the first one or two letters than will shorter words, and this factor will tend to act against the discovery of an effect of letter length. At present, we do not know whether number of letters or horizontal extent underlies any length effect in spatial neglect dyslexia. This issue will only be resolved by studies which systematically and independently vary the two factors.
Another difference between A.H. and V.B. which may be more apparent than real concerns their performance with real words like SOFA as compared to invented nonwords like BOFA. A.H. showed a dramatic difference in performance between words and nonwords, reading 42% of words accurately in one test as compared with only 3% of nonwords. In contrast, V.B.’s accuracy with nonwords (87.6% correct) was comparable with her performance with words. As with length effects, however, we think that there are good reasons for not jumping straight to the conclusion that A.H.’s and V.B.’s neglect dyslexias were different in some important way. First, V.B.’s reading performance was overall much better than A.H.’s, with V.B. performing at over 85% correct with words as compared with A.H.’s level of around 40–45%. Ellis et al. (1987a) presented data and arguments to suggest that V.B. must have been perceiving many words correctly in their entirety, and that her neglect dyslexia only affected a minority of her attempts at single words. If this is so, then she may also have perceived many nonwords correctly in their entirety.
Comparisons between words and nonwords are also confounded by differences in “guessability”. Suppose that a patient is asked to read SOFA, but because of neglect dyslexia only creates a visual representation that is something like -OFA. Provided that the patient knows in advance that the target is a word, the item should be read correctly (because there is only one four-letter word ending in -OFA). In contrast, if the target is BOFA, then even if the patient knows that a nonword response is required, it is unlikely to be guessed correctly (because -OFA as a nonword could also be DOFA, NOFA, WOFA, etc.).
A patient like A.H., whose neglect dyslexia is severe, may perceive few or no initial letters correctly, and may be more prone to differences in guessability than a patient with milder neglect dyslexia like V.B. Over and above that is the fact that the potential for differences in guessability to engender apparent lexicality (word-nonword) effects will depend crucially upon the selection of items. Instead of SOFA and BOFA, we could have contrasted PULL with VULL. There are roughly as many four-letter words beginning with a consonant and ending -ULL as there are potential nonwords, so guessability would not favour words here. Overall, though, guessability would favour words over nonwords in patients who knew in advance whether the target was a word or a nonword and were clear about the distinction in the first place. A proper comparison taking all these factors into account has yet to be carried out, so a final verdict as to the reality of word-nonword differences in neglect dyslexia cannot be formed.
Word Errors, Letter Errors and Hemianopia
V.B. (Ellis et al., 1987a) omitted words from the beginnings of lines of text. Her misreadings involved a mixture of neglect and visual errors, and her neglect errors involved a mixture of different sorts (e.g. letter deletions vs letter substitutions). There are reasons, though, to think that different mechanisms and deficits underlie these different categories of error.
Young, Newcombe and Ellis (1991) reported a patient, S.P., who resembled V.B. in several respects. In text reading, she omitted the initial words of lines, and she made neglect dyslexic errors to single words (e.g. JANGLE tangle, MARROW arrow). In fact, S.P. misread more words given unlimited exposure times than did V.B. (29% errors for S.P. compared with around 8% for V.B.), and a higher proportion of her errors met Ellis and co-workers’ (1987a) criterion for being counted as neglect errors. Performance improved when words were presented vertically, implying that, as with V.B., S.P.’s neglect dyslexia was determined by the spatial location of letters in words.
In an investigation of whole word omissions vs misreadings, S.P. was presented with sheets containing 15 words positioned semi-randomly in an array made from five columns and 15 rows, but with only one word per row and three per column (see Young et al., 1991, figure 3). Across a series of sheets, she was presented with 315 words, with 63 words in each of the five vertical columns. The words were of 4 to 6 letters and shared the common property that the first letter can be either deleted or substituted to produce another words (e.g. FLAP, the F of which can be deleted to lap or substituted to slap). In this task, S.P. made 77 omission errors (complete failures to attempt words) and 48 misreadings, 47 of which were clear neglect errors. Whole-word omissions and neglect dyslexic misreadings showed very different patterns of distribution across the five columns: whole-word omissions were concentrated in the left-most columns, whereas misreadings were distributed much more evenly across the columns. These findings suggest that omissions and misreadings may have somewhat different underlying causes.
Patient J.O.H. of Costello and Warrington (1987) made neglect dyslexic errors on the initial letters of words. He also made errors of substitution, omission and addition to whole words in text, but we are told that these word errors “occurred both to the left and the right and the centre of the page. On no occasion did he fail to ‘read’ to the end of a line or return to the beginning of the next line. There was no evidence of any unilateral spatial neglect in his reading of prose” (Costello & Warrington, 1987, p. 1114). In contrast, patient V.S.N. of Kartsounis and Warrington (1989) showed severe neglect when attempting paragraphs of text, but minimal neglect dyslexia to single words. When asked to read normal passages of text, he typically failed to attempt more than three words from the extreme right of each line, but he read 94 of a set of 96 single words correctly (only misreading DAYBREAK as “tapebreak” and WAYSIDE as “quayside”).
There would seem to be a distinction to be drawn between the form of unilateral neglect that causes patients to omit whole words at the beginnings of lines of text and the form of neglect that causes neglect dyslexic misreadings. Another distinction to be drawn is between those misreadings which affect the initial letters only (neglect errors) and misreadings which affect other portions of words (visual errors). Examples of clear visual errors from V.B. include DIVINE living, AISLE praise and MARQUEE quarrel. We have mentioned already that S.P. misread more words than V.B., but that a higher proportion of her errors were neglect errors. This is because V.B. made a higher proportion of visual errors than S.P. If we consider only real-word errors, and apply Ellis et al. criterion for what is to count as a neglect error, then 94% of S.P.’s errors were neglect errors, whereas the corresponding figure for V.B. was 72%. Ellis et al. (1987a) very strict criterion almost certainly rejects some genuine neglect errors on the grounds of shared letters to the right of the putative neglect point in the target and error words (e.g. INSIDE fireside, ELBOW brow). Nevertheless, if visual and neglect misreadings arose from a common underlying deficit, it might be difficult to account for two patients, one of whom (S.P.) makes more errors overall than the other (V.B.) while making a dramatically higher proportion of neglect errors.
All that remains in this section is to compare neglect dyslexic misreadings which involve omissions of initial letters (e.g. LEVER ever, DEARTH earth) with misreadings in which initial letters are substituted (e.g. RIVER liver, SABLE table). We shall argue that which type of error occurs may depend upon a subtle interaction between neglect dyslexia and another common symptom in patients with neglect dyslexia, namely hemianopia.
Like many patients with neglect, V.B. had a left homonymous hemianopia (blindness to stimuli in the left visual field). Many patients with hemianopia do not show neglect dyslexia, and three of the six neglect dyslexic patients studied by Kinsbourne and Warrington (1962) did not have hemianopias, suggesting that hemianopia cannot be held solely responsible for spatial neglect dyslexia. In addition, Kinsbourne and Warrington (1962), Ellis et al. (1987a) and Young et al. (1991) have all done experiments to show that neglect errors can occur when words are presented entirely within the intact right visual field. Nevertheless, we propose that neglect dyslexia and hemianopia may interact to determine the precise error pattern observed in a particular patient.
Young et al. (1991) contrasted the effect of displaying words entirely within S.P.’s intact right visual field with the effect of presenting words centrally, so that they straddled the boundary between S.P.’s intact right visual field and her hemianopic left visual field. Presentation times were unlimited, so that S.P. was free to move her eyes; what varied was the place at which the word initially appeared. With presentation initially in the right visual field, the majority of S.P.’s errors involved substitutions of initial letters (e.g. BAT eat, BELIEF relief). With central presentation, a higher proportion of errors involving the simple omission of initial letters was observed (e.g. CAT at, CHAIR hair, VALLEY alley).
We propose that if a word fell entirely within S.P.’s right visual field, then all letters were processed to some degree, and the implicit awareness of the existence of initial letters manifested itself in errors involving substitution rather than simple deletion of initial letters. If a word straddled the boundary between the hemianopic left visual field and the intact right visual field, and if a word was embedded within the right portion of the target word, then the reluctance to scan in the leftwards direction that is known to characterise patients with neglect caused S.P. to stop once she had located a whole word, the result being an omission error. So, for example, if the word CHAIR fell entirely within S.P.’s right visual field, she might read it correctly or make a substitution error (e.g. flair), but if it straddled the boundary between her hemianopic left visual field and her intact right visual field, then she might scan leftwards only until she saw AIR or HAIR and so make an omission.
Thus, it would appear that hemianopia does not cause spatial neglect dyslexia, but may interact with it to determine the observed proportions of letter deletions vs letter substitutions. Words falling entirely within intact areas of the visual field are likely to be misread by having their initial letters substituted, but letters may be deleted from words falling across the boundary between intact and hemianopic areas of the visual field (especially if the right end forms a familiar word). In sum, we would argue that different underlying deficits are responsible for whole-word omissions, visual errors and neglect errors in spatial neglect dyslexia, and that neglect dyslexia alone (unhindered by interactions with hemianopia) typically results in substitutions rather than deletions of initial letters.
Attention and Spatial Neglect Dyslexia
Riddoch, Humphreys, Cleton and Fery (1991) described a patient, J.B., with a pattern similar to that of both V.B. and S.P. J.B. showed left-sided neglect on a range of tasks including reading following a right temporo-parietal injury. Typical neglect dyslexic errors occurred (e.g. GROSS cross, BOUGH slough), with 63% of target-error pairs sharing the same number of letters (indicating a tendency to substitute rather than add or omit initial letters). When words were rotated through 180° so that they were upside-down with their initial letters on the right, J.B. read 91% of words correctly (as good a level of performance as he attained with letters in conventional formats). Fifteen of the 20 errors made to upside-down words involved the left-most (final) letter, with 3 involving central letters and only 2 involving the right-most (initial) letter.
In an earlier study, Riddoch and Humphreys (1983) had sought to test the hypothesis that unilateral neglect is due to a failure to attend to the left sides of objects by showing that neglect is reduced if the patient is required to locate and identify a cue presented to the left of the object. In their experiment, they required neglect patients to bisect horizontal lines with or without a digit to the left of the line. Neglect patients tend to bisect to the right of true centre, presumably because of a failure to respond to the leftmost portion of the line. When the digit was present, and had to be reported before the patient bisected the line, this tendency was reduced.
Rather than using digits, Riddoch et al. (1991) placed a hash sign (#) to the left of words that J.B. was to read aloud. Performance improved relative to his reading of the same words on a previous occasion though, as Riddoch et al. admit, their failure to compare reading of words with and without cueing in the same sessions means that the cueing effect might be due to improvement over time in J.B.’s general performance. That said, we can report data on V.B. that seems at least compatible with an attentional account of neglect dyslexia. V.B. was asked to read five-letter words printed in upper-case (capital) letters. The words were printed with either the first two letters in bold and the last three not (e.g. BROOK) or the last three bold and the first two not (e.g., SABLE. The reasoning was that the letters in bold would be more salient and would attract attention more than those letters not in a bold face. In four different within-session comparisons, V.B.’s scores out of 40 for words with bold initial vs non-bold initial letters were 31 vs 25, 36 vs 33, 31 vs 26 and 31 vs 26. It would thus seem that increasing the perceptual salience of initial letters improves reading performance in left-sided spatial neglect dyslexia, as an attentional theory would predict.
To summarise, we shall reserve the term “spatial neglect dyslexia” for the tendency to misread the initial letters of words in conventional formats, but the final letters of words presented upside-down or mirror-reversed (i.e. a tendency to misread the left-most letters, however the word is presented). When spatial neglect dyslexia is not complicated by the presence of a hemianopia, the dominant error would appear to be the substitution (rather than the omission) of left-most letters, with the result that target and error words tend to be of similar lengths. Comprehension matches the misreading rather than the target. Words which occupy a greater horizontal extent are more likely to be misread, but it is not yet clear whether an effect of number of letters exists over and above an effect of physical length. Words displayed in vertical format are read much more accurately than words in normal horizontal format. The tendency to omit words at the beginnings of lines of text would appear to be a dissociable symptom with a separate cause, though it is a symptom which often accompanies spatial neglect dyslexia. Visual misreading errors may reflect a third, commonly co-occurring deficit. Finally, the effects of cueing and perceptual salience of initial letters are compatible with the hypothesis that spatial neglect dyslexia results from inattention to left-most letters in strings.
Neglect and Reading II: Positional Neglect Dyslexia
In contrast to the patients just discussed, whose errors affect the left-most letters of words, patients have been reported whose misreadings affect the final letters of words. As we shall see, their condition cannot be explained as simply the mirror image of left-sided spatial neglect dyslexia. Warrington and Zangwill (1957) reported a patient who made errors on final letters, misreading BEWARE as because, TONGUE as together, and OBTAINED as oblong. A detailed report of another patient, N.G., who made final letter errors has been provided by Hillis and Caramazza (1989; 1991; Caramazza & Hillis, 1991). N.G. was naturally left-handed, though she had been taught to write with her right hand, and had a left hemisphere lesion. N.G. was not aphasic, nor did her short-term memory appear to be seriously impaired (forward span 7; backward span 3). Although she had shown neglect on a variety of tasks in the period immediately following her stroke, by the time that the data discussed here were collected, neglect was only evident in her processing of written language.
When attempting to read sentences, N.G. would occasionally fail to attempt to read the right-most words. Among the words she attempted would be errors focused on the end letters. Thus she read “The quick brown fox jumps over the lazy dog” as “The quiet brown fox jumped over the lazy doctor”. She read around 75% of single words correctly, with her errors tending to involve the final letters (e.g. PARK part, HUMID human, HOUND house, SPRINTER sprinkle). As with V.B., target and error words tended to be of the same length, and comprehension matched the error word rather than the target.
In contrast to patients with spatial neglect dyslexia, N.G. could name all the letters in a word correctly, yet would still misread (and misconstrue) the word. For example, when asked to name the letters in JOURNAL and then read the word, she responded “J, O, U, R, N, A, L, … journey”. This contrasts with V.B. responding to BARRIER as “H, A, R, R, I, E, R … harrier” and to CANISTER as “B, A, N, I, S, T, E, R … banister”. Asked to name the letters of STATUS and define the word, N.G. responded, ‘S, T, A, T, U, S … The Statue of Liberty is very pretty … The word is statue”. To PLANET she responded, “P, L, A, N, E, T … The plane landed on the other side … The word is plane”.
N.G.’s ability to name the component letters of words she went on to misread shows that her deficit cannot have lain within the earliest stages of visual letter identification. There are other aspects of N.G.’s performance that also suggest a more central locus for her neglect dyslexia. Patients with spatial neglect dyslexia make errors to left-most letters, and so misread the initial letters of ordinary words but the final letters of words presented upside-down or mirror-reversed. With words presented vertically, their errors are no longer concentrated in any particular portion of words. N.G. is different in that she continued to make errors on the final letters of words whatever their orientation. Errors to mirror-reversed words include COMMON comet, REGULATED regular and DISCOVERY disco, while errors to words presented vertically include RANG ran, BLENDING blemish and MOTIONLESS motel. Because N.G.’s errors occur at the ends of words irrespective of their orientation, her neglect dyslexia might be termed positional neglect dyslexia. (The term “positional” is taken from Katz and Sevush (1989), whose patient J.M. is claimed to have made errors on initial letters of words irrespective of their orientation, though the evidence upon which that claim is based is less compelling than for N.G.)
Whereas the ability of spatial neglect dyslexics like V.B. to identify words spelled aloud to them can be very good, N.G. made errors on this task that had the same properties as her errors to written words. Examples include “B, A, S, I, S” brass, “S, P, A, R, R, O, W” space and “E, A, R, N, S” earring. These errors suggest that N.G.’s deficit lies at a stage beyond the point at which input from spoken letter names converges with that from written words. N.G.’s reading was also unaffected by extending the horizontal length of words by inserting spaces between letters. This again suggests that N.G.’s deficit is of a less spatial nature than that of V.B. and the other spatial neglect dyslexics.
Ellis et al. (1987a) noted that, for target words of a wide range of different lengths, V.B.’s neglect point fell most frequently after the first letter (from FAT sat to POPULATION copulation). Caramazza and Hillis (1991) analysed N.G.’s reading errors and showed that “(1) the vast predominance of errors occurred on the right half of words irrespective of their length; (2) errors occurred at equal rates as a function of absolute distance from the centre of a word; and (3) errors increased ‘linearly’ as a function of distance from the centre of the word” (p. 406). The evidence for this claim is based on an analysis of N.G.’s errors in which errors were scored at different positions in words ranging in length from four to nine letters. When the results are displayed as in Table 11.1, showing the percentage of errors at each position with the centres of different word lengths aligned, the validity of Caramazza and Hillis’s claims can be seen. (Note that this is not the pattern one would expect to see if a short-term memory problem of any sort was preventing N.G. from “holding onto” early letters in words when processing later ones.)
Table 11.1
Percentages of Reading Errors of All Types as Function of Letter Position in Words of Different Lengths for Patient N.G.a
aFrom Caramazza and Hillis (1991). Reproduced with permission, n = number of words presented of a given length.
Table 11.2 presents the results of a similar analysis of V.B.’s errors. For this analysis, an error was credited to a given position in a word if the letter in that position was omitted or substituted. Additions to the beginning of words (e.g. TOP stop) were scored as errors at the initial position. Table 11.1, taken from Caramazza and Hillis (1991), combines N.G.’s neglect and visual errors involving the production of both word and nonword responses to word targets. We discussed earlier the possibility that in spatial neglect dyslexia visual and neglect errors have a different basis. The analysis in Table 11.2 is limited to a corpus of V.B.’s real-word neglect errors (an augmented version of the errors given in the appendix to Ellis et al., 1987a). The restriction to neglect errors, plus the fact that V.B.’s overall error rate was less than N.G.’s, means that the percentages are lower than in Table 11.1 (and are based on smaller numbers). Nevertheless, it can be seen that V.B.’s errors tend to afflict the right halves of words. Thus, although the first position was most error-prone for all lengths, some errors were recorded affecting the first four letters of eight-letter words, whereas only the first two letters of three- and four-letter words were affected (and, in fact, only a single error was recorded at the second positions of three- and four-letter targets).
Table 11.2
Percentages of Real-word Neglect Dyslexic Reading Errors as a Function of Letter Position in Words of Different Lengths for Patient V.B.a
aBased on a supplemented version of the corpus presented in Ellis et al. (1987a). n = number of words presented of a given length.
V.B.’s errors to words in normal horizontal format are restricted to the left halves of words, whereas for N.G. the restriction is, of course, to the right halves. It is also true for V.B. as for N.G. that errors increase with distance from the centre. What is not clear is that V.B.’s errors occur at equal rates as a function of absolute distance from the centre of a word, as N.G.’s did. This claim for N.G. is based on the fact that if one scans vertically down the right half of Table 11.1, the error rates look very similar. Thus, the error rates are comparable at the fifth position of five-letter words (18%), the sixth position of seven-letter words (16%) and the seventh position of nine-letter words (23%). That is not true for V.B: for example, the error rates are not comparable at the first position of four-letter words (9%), the second position of six-letter words (3%) and the third position of eight-letter words (2%). We should be cautious in interpreting this observation: Tables 11.1 and 11.2 are based on different error types (all errors for N.G.; real-word neglect errors only for V.B.). Also, N.G. was presented with typewritten words, whereas handwritten words were shown to V.B. and, as noted earlier, there may have been some tendency to space out the letters of short words more than long words. But there is a possible difference here between spatial and positional dyslexia that warrants further investigation.
A possible refutation of the conjecture that all neglect dyslexias are word-centred is provided by patient T.B. of Patterson and Wilson (1991). T.B.’s errors were confined to the initial letter of a word, irrespective of its length. Errors were distributed equally throughout words presented vertically and tended to affect the ends of mirror-reversed words. These observations, plus the fact that his oral spelling and identification of words spelled aloud were both excellent, suggest that T.B.’s neglect dyslexia was spatial rather than positional in nature (though see Caramazza & Hillis, 1991, for discussion).
Patient R.H.R.
Warrington’s (1991) patient, R.H.R., was a right-handed man with a small left hemisphere lesion. There was no suggestion of perceptual impairment or unilateral neglect evident in his drawing or cancellation. His speech was fluent without apparent word-finding difficulties, but his reading and spelling were both impaired.
R.H.R. read 55% of a set of common words correctly, and his errors were concentrated towards the ends of words. Fifty-three per cent of his errors to words presented horizontally satisfied Ellis and co-workers’ (1987a) criterion for neglect errors except that, being right-sided errors, the target and error words were identical to the left of an identifiable neglect point in each word and shared no letters in common to the right of the two neglect points. Examples are CALL calf, TRUTH truck and SEASON seaside. The remaining errors were mostly visual errors (e.g. TOOK talk, AGREE argue) or failures to respond.
With vertical presentation of words, R.H.R.’s accuracy fell to 13% correct, and he now made more visual errors than neglect errors. Nevertheless, 25% of his errors continued to satisfy the strict criterion for what should count as a neglect error and, importantly, remained focused on the final letter(s) of the target words. Examples include KING kite, MARY march, BLOW blue and COAT coach. The fact that R.H.R. continued to make errors on end letters when words were presented vertically, suggests that he should be regarded as a case of positional rather than spatial neglect dyslexia.
Warrington (1991) provides little information on R.H.R.’s spelling, but N.G. showed features which mirrored in a remarkable way her positional neglect dyslexia. But before considering the nature and locus of the impairments in spatial and positional neglect dyslexia, it is helpful to review the impact of unilateral neglect on the production of writing.
Spelling and Writing and Unilateral Neglect
We have seen that N.G. made errors on the initial letters of words whether they were presented aurally (by the experimenter spelling the word aloud) or visually, and irrespective of the orientation of written words. N.G.’s spelling and writing showed features which mirrored in a remarkable way her positional neglect dyslexia. In both spelling aloud and writing, N.G. made errors which were again concentrated at the ends of words. These errors typically resulted in the production of nonwords. Examples of her errors in spelling words aloud to dictation include sneeze S, N, E, E, D, taught T, A, U, G., H, T, H and events E, V, E, N, I, S. Examples from writing to dictation include floor FLOORE, jury JURD and avoid AVOILOE. N.G. was more accurate at writing words (33% correct) than nonwords (21% correct), but her spelling or writing of nonwords to dictation showed the same concentration of errors at the ends (e.g. skart SKARR, remmun REMNEY, chench CHEN).
N.G.’s spelling accuracy declined with increasing word length (from 50% correct on four-letter words to 7% correct on eight-letter words). When asked to spell words aloud backwards, she still made errors on the final letters, even though those letters were now the first to be spoken. Thus, asked to spell “absorb” aloud backwards she produced N, W, O, S, B, A (= ABSOWN forwards), and asked to spell “oyster” aloud backwards she produced E, T, S, Y, O ( = OYSTE). This demonstrates that her spelling/ writing deficit was, like her reading deficit, positional rather than spatial. An analysis similar to that carried out on her reading errors showed that mistakes in spelling and writing also occurred in the right halves of words, whatever their length.
The situation for the patients with spatial neglect dyslexia is different. V.B.’s oral spelling was excellent, as was her identification of words spelled aloud to her, but her attempts at writing contained numerous errors (Ellis, Young, & Flude, 1987b). When writing text, V.B. wrote over towards the right side of the sheet of paper, leaving a wide left margin. In her handwriting, she often failed to go back in order to dot i’s and cross t’s. Finally, her attempts to write individual words contained frequent errors involving the omission or repetition of letters or letter strokes. J.B. showed the same pattern (Riddoch, 1991), but Ellis et al. (1987b) mention that patient S.P., whose neglect dyslexia is analysed by Young et al. (1991), showed V.B.’s tendencies to write on the right side of the page and to fail to dot i’s and cross t’s, but did not omit or repeat letters or strokes.
It may be that writing over to the right of the page and failing to dot i’s and cross t’s reflects the more general visuo-spatial neglect shown by many of the patients with spatial neglect dyslexia. It is obvious why visuo-spatial neglect could cause a patient to ignore the left side of the page in writing. Errors on i’s and t’s may arise because going back at the end of a word to dot the i’s and cross the t’s means moving leftwards into the neglected region of space, something which patients with visuo-spatial neglect have difficulty doing. The omissions and repetitions of strokes and letters occurred in V.B. and J.B. but not S.P. Lebrun (1976) used the term “afferent dysgraphia” to refer to this pattern, noting the similarity of these errors to those made by normal subjects writing with their eyes closed, and arguing that letter and stroke omissions and repetitions arise when a patient or a normal writer is unable to use visual and kinaesthetic feedback to monitor and control the complex sequences of movements required in handwriting. Ellis et al. (1987b) asked normal subjects to write with their eyes closed (disrupting visual feedback) and while tapping a finger sequence with the non-writing hand (to interfere with the monitoring of kinaesthetic feedback). They showed that under these conditions normal subjects made letter and stroke repetition errors of the same type and with the same frequency as shown by V.B. writing under normal conditions. It may be that monitoring visual and kinaesthetic feedback is a quasi-attentional function of the normal right parietal lobe which is not directly related to unilateral neglect (because letter and stroke errors do not occur towards one side of words), but which is likely to co-occur with aspects of unilateral neglect as a result of the anatomical proximity of the regions that mediate these functions. Behrmann and co-workers’ (1990) report that A.H. showed a spatial neglect dyslexia without afferent dysgraphia supports the proposal that different deficits underlie these two symptom patterns.
Thus, spatial neglect dyslexia appears not to be linked to dysgraphia in any necessary way, but in the one clear case of positional neglect dyslexia reported to date, N.G., there existed a complementarity between the characteristics of reading and writing that looks to be more than coincidental. The question then arises as to whether there exist neglect-related disorders that are specific to spelling and writing and which can occur in the absence of neglect dyslexia.
Caramazza, Miceli, Villa and Romani (1987) reported a patient, L.B., who made letter omission and substitution errors in both writing and spelling aloud, and for both words and nonwords. Reading of familiar words was preserved, though reading of unfamiliar nonwords was impaired (Caramazza, Miceli, Silveri, & Laudanna, 1985). Like normal writers (Wing & Baddeley, 1980), L.B.’s errors were concentrated in the middles of words. Hillis and Caramazza (1989) report two patients, D.H. and M.L., whose errors resembled L.B.’s, involving letter omissions and substitutions, but M.L.’s errors showed a shift towards the beginnings of words, whereas D.H.’s errors showed a shift towards the ends. In neither case were the beginning or final letters most affected; rather, there was a shift in error focus away from the centre towards the initial (M.L.) or final (D.H.) letters.
One caveat should be entered here. Our knowledge of the distribution of normal errors is derived from a study (Wing & Baddeley, 1980) which combined errors from many different writers. The distribution they obtained, which more or less matches L.B.’s distribution, is a distribution for the population of reasonably skilled writers of English. It may not accurately represent the error distribution of each individual normal writer. It is at least conceivable that there are normal writers whose error distributions are, like those of D.H. and M.L., shifted away from centre. Hence we cannot rule out the possibility that D.H.’s and M.L.’s spelling errors showed similar distributions pre-morbidly, and that the effect of their brain injury has been only to increase their overall tendency to error. Whether or not this is the case, the general point we wish to make is that a discipline like cognitive neuropsychology which pays such close attention to individual differences among patients should also take careful cognisance of individual differences among normal subjects. When comparing patient data against normal data, and especially when differences between patients of similar types are being considered (as with L.B., D.H. and M.L.), it is vital that the normal data should incorporate measures of range and variation, not just central tendency.
That said, the shift in M.L.’s spelling and writing errors towards the beginnings of words was accompanied by a left-side visuo-spatial neglect and very mild neglect dyslexia, while the shift in D.H.’s spelling and writing errors towards the ends of words was accompanied by a mild left-side visuo-spatial neglect and reading errors concentrated on the ends of words. Hence in both cases, the neglect impairment indicated by a shift in spelling and writing errors was not entirely specific, being associated with signs of similarly lateralised neglect in other tasks.
Baxter and Warrington (1983; 1990) describe a left-handed patient, O.R.F., with a right parietal lesion who had a mild fluent aphasia and only a slight left-sided neglect dyslexia. O.R.F. was unable to use his preferred left hand to write because of a left hemiplegia, and his attempts to write using his right hand are described as “extremely slow and hard to decipher”. He could, however, attempt to spell aloud. When he did so, his accuracy was affected by the length of words to be spelled, but not by their frequency in the language or by their spelling-sound regularity. The errors he made were concentrated at the beginnings of words. This was true whether he spelled words aloud forwards or backwards.
Baxter and Warrington (1983) suggest that O.R.F. “appeared to be spelling from an ‘inner screen’” and that a left-sided neglect reduced the efficiency of this process. Bisiach and co-workers have shown that the report of mental images can be affected by unilateral neglect (Bisiach & Luzzatti, 1978), but as far as we are aware, every patient who has been observed to show neglect of mental images has also shown visuo-spatial neglect when dealing with external visual stimuli (e.g. in drawing or cancellation). O.R.F. is reported as not showing such visuo-spatial neglect for external stimuli or even for non-verbal mental images (D. Baxter, pers. comm.); hence he would not seem to represent a case of imagery neglect without visuo-spatial neglect, but rather a selective neglect dysgraphia caused by a disruption to orthographic representations involved in spelling but not in reading.
Neglect and Visual Language
Theoretical accounts of the psychological processes underlying reading and spelling/writing differ considerably in their particulars but at the most general level there is broad agreement. A word on a page is a pattern of lines and curves which must undergo processing if it is to be recognised, comprehended and pronounced successfully. First, at least some of the component letters of the word must be identified and the position of letters in the word must be noted. Letter position is important because words may contain the same letters and only differ in their positions (e.g. SLAP/PALS/ ALPS/LAPS).
Familiar words come in a wide range of fonts, sizes and formats, and in either UPPER-CASE or lower-case. One way for the reading system to be able to recognise all the different possible versions of a familiar word would be to have each letter string undergo some “normalisation” process resulting in the creation of a positionally encoded sequence of abstract letter identities (Coltheart, 1981) or graphemes (Caramazza & Hillis, 1991). If, following such normalisation, a string of graphemes is to be recognised as a familiar word, it must be matched against stored representations of all the words known to the reader. The memory structure in which these stored representations are held goes under various names, including the visual input lexicon and the orthographic lexicon.
Spatial neglect dyslexia affects the left side of both words and nonwords. It is influenced by physical length and salience of letters. Errors to vertical words are randomly dispersed and identification of words presented as oral spellings can be preserved. This pattern is taken by Caramazza and Hillis to indicate that the neglect responsible for spatial neglect dyslexia affects an early stage of visual processing, and an early representation: one which precedes the formation of an abstract graphemic representation. Caramazza and Hillis (1991) suggest that the representation in question is a “stimulus-centred letter shape map” based on left-right not first-last coding.
When word recognition is not affected by hemianopia, the natural error type in spatial neglect dyslexia appears to be the substitution rather than simple deletion of initial letters. Those letters must be processed sufficiently for their presence to be covertly acknowledged in the error. Ellis et al. (1987a) proposed that spatial neglect dyslexia affects the encoding of the identity of letters on the left of strings, but that positional encoding is relatively preserved. Hence, spatial neglect dyslexia might cause a word like FACE to be converted into an abstract graphemic representation that is something like < - > 1, < A > 2, < C > 3, < E > 4, where the identity of the left-most letter has not been accurately encoded, but its presence has, with the result that the A, C and E are encoded as the second, third and fourth letters, rather than the first, second and third. This representation would activate the entries in the orthographic lexicon of words like LACE and PACE more than it would activate the representation for ACE, so that a misreading is more likely to involve a substitution of the first letter than its deletion.
Positional neglect dyslexia as displayed by N.G. affects final letters irrespective of their spatial positioning, and affects the identification of words from oral spellings as well as in written form. It is not influenced by physical length (nor, we would predict, by manipulations of letter salience). Caramazza and Hillis (1991) note that the characteristics of positional neglect dyslexia are compatible with the suggestion that positional neglect dyslexia affects the abstract graphemic representations themselves (rather than the processes which create the abstract graphemic representations, as in spatial neglect dyslexia).
Cognitive and neuropsychological accounts of the processes underlying spelling and writing propose that the spellings of familiar words are also retrieved from an orthographic lexicon in the form of abstract letter identities or graphemes which are subsequently translated through a series of intermediate stages either into letter names (in spelling aloud), or into arm, hand and finger movements (in handwriting or typing) (see Ellis, 1982; 1988). N.G.’s oral spelling and writing showed word-final errors whose distribution matched that of her reading errors in a remarkably close way, suggesting that the same (disrupted) representation may have been involved in reading, spelling and writing. Caramazza and Hillis (1991) propose that a word-centred, positional, abstract graphemic representation serves both as the input to the orthographic lexicon in reading and as the initial representation of words retrieved from the lexicon in spelling and writing. We have discussed logical reasons why visual word recognition might employ abstract representations (to do with generalising over different formats, fonts and cases). There are also empirical lines of evidence commensurate with this view (see, e.g. Carr, Brown, & Charalambous, 1989; Rayner, McConkie, & Zola, 1980).
Hillis and Caramazza’s analysis of N.G. presents the first evidence we are aware of suggesting that a common abstract graphemic representation may be involved in both reading and spelling/writing. A single abstract graphemic representation could, in principle, interface with separate orthographic lexicons for input and output, but the architecture in that case would be somewhat strange. The question of whether reading and spelling/writing are subserved by one lexicon has been much debated but not resolved (see Coltheart & Funnell, 1987; Monsell, 1987), so the idea of a common input-output graphemic representation interfacing with a common input-output orthographic lexicon remains very much within the bounds of possibility.
N.G. could name the letters of words she went on to misread. It could be that spelling aloud involves treating each of the component letters of words as objects, and that her neglect does not affect single objects, showing itself only when object strings are being processed. Finally, N.G. showed the same abstract positional deficit in the reading and writing of unfamiliar nonwords as well as familiar words. There is current debate over whether nonwords are read and written through the intercession of processes distinct from those mediating word reading and spelling, or whether lexical processes can also cope with nonwords (see Patterson & Coltheart, 1987, for a review of the arguments and evidence). Either way, the abstract graphemic representation disrupted in N.G. must be common to both lexical and sublexical processing as well as to both reading and spelling/writing.
Certain architectures for cognition, notably parallel distributed processing systems (Quinlan, 1991), do not (explicitly) honour the distinction between graphemic representations and the lexical units or nodes which recognise or generate them. Within such systems, a familiar word is a pattern engraved across finer grain representations. It could be that N.G.’s positional neglect dyslexia attacks a representation which serves as both input to and output from the orthographic lexicon(s), but we think it at least worth contemplating the possibility that the representation which is disrupted is the orthographic lexicon itself. This would imply that such central representations require their activation levels to be sustained in a manner similar to that by which spatial representations of the external world are sustained.
At this point, arbitrating between rival hypotheses becomes increasingly difficult in the absence of explicit simulations of the processes in question. Mozer and Behrmann (1990) have created a computer model which attempts to show how attentional processes could modulate word recognition, and how “lesioning” such a model could result in phenomena like those seen in spatial neglect dyslexia. Among other things, it provides an account of word omissions and of the superior reading of words than nonwords, though it does not (yet) provide a natural account of other effects, such as the spatial/positional distinction and the dominance of letter substitution over deletion errors. It is early days in the evolution of such models, but the enterprise is clearly a valuable one. We can only learn from attempts to translate sometimes rather vaguely expressed ideas about attention and neglect into working simulations.
Analysis of spatial and positional neglect dyslexias indicates that unilateral neglect can affect at least two separate stages or representations involved in reading and spelling/writing. The studies of patients with neglect-like disorders of writing suggest that still other spatial representations involved in processing visual language may be prone to unilateral neglect.
The letter omissions and substitution errors of Hillis and Caramazza’s (1989) patient M.L. tended to occur at the beginnings of words, while the errors of patient D.H. tended to occur at the ends. M.L. also showed a leftside visuo-spatial neglect and (very mild) neglect dyslexia, while D.H.’s spelling and writing errors towards the ends of words were accompanied by a mild left-side visuo-spatial neglect and reading errors concentrated on the ends of words. It is possible that M.L. and D.H. showed positional neglect dyslexias much milder than N.G.’s, resulting in low levels of reading errors and slight shifts in distributions of spelling and writing errors. If so, then M.L. is the first reported case of left-sided positional neglect dyslexia; if not, then as with patient O.R.F. (Baxter & Warrington, 1983; 1990), their disorders will need to be located downstream of the abstract graphemic level affected in N.G. And, if not, future research will need to explore whether, in patients with left hemisphere language dominance, spatial neglect dyslexia is invariably left-sided, resulting from right hemisphere injury, and positional neglect dyslexia invariably right-sided, resulting from left hemisphere injury. Such a state of affairs could arise if the right hemisphere is responsible for maintaining attention across relatively peripheral visual representations of the external world, while the left hemisphere sustains attention (or activation) over more central, but still positionally coded, linguistic representations.
Most patient with neglect dyslexia also show visuo-spatial neglect, including patient N.G. whose positional neglect dyslexia appeared to involve fairly abstract representations (Caramazza & Hillis, 1991; Hillis & Caramazza, 1989; 1991). There are signs, though, that visuo-spatial neglect and neglect dyslexia may dissociate. Patient J.O.H., reported by Costello and Warrington (1987), made errors on the initial letters of words (e.g. MOWER shower, RAIN pain), but in visuo-spatial tasks like drawing, cancellation and line bisection he showed signs of right-sided neglect. J.O.H. would seem to represent in one individual a double dissociation between a left-sided neglect dyslexia and a right-sided visuo-spatial neglect. Unfortunately, Costello and Warrington’s (1987) description of J.O.H. does not provide us with the data that would allow us to confirm that J.O.H.’s neglect dyslexia was spatial in nature. No cases of left-sided positional neglect dyslexia have yet been reported, but it is conceivable (J.O.H. was an unusual patient) that J.O.H. may have represented a case of left-sided positional neglect dyslexia (i.e. the mirror-image of N.G.). If that were the case, conclusions drawn from J.O.H. would not apply automatically to patients with spatial neglect dyslexia. The point is that if, as we now know, neglect dyslexia takes more than one form, then the dissociation of each form with other manifestations of neglect must be established separately. In order to make theoretical progress, we need to know for the left- and right-sided forms of each variety of neglect dyslexia the extent to which that variety can dissociate from neglect in other modalities and for other types of material. If unilateral neglect can affect representations involved in the production of written language, as the study by Hillis and Caramazza (1989) suggests, then we also need to know more about other manifestations of “motor neglect” (Laplane & Degos, 1983) in such patients.
References
Assal, G. & Zander, E. (1969). Rappel de la symptomologie neuropsychologique des lésions hémisphériques droites. Archives Suisses de Neurologie Neurochirurgie et de Psychiatrie, 705, 217–239.
Baxter, D.M. & Warrington, E.K. (1983). Neglect dysgraphia. Journal of Neurology, Neurosurgery and Psychiatry, 46, 1073–1078.
Baxter, D.M. & Warrington, E.K. (1990). A comment on Hillis and Caramazza’s (1989) paper “The graphemic buffer and attentional mechanisms”. Brain and Language, 38, 615–616.
Behrmann, M., Moscovitch, M., Black, S.E., & Mozer, M.C. (1990). Perceptual and conceptual mechanisms in neglect dyslexia. Brain, 113, 1163–1183.
Bisiach, E. & Luzzatti, C. (1978). Unilateral neglect of representational space. Cortex, 14, 129–133.
Caramazza, A. & Hillis, A.E. (1991). Levels of representation, co-ordinate frames, and unilateral neglect. Cognitive Neuropsychology, 7, 391–445.
Caramazza, A., Miceli, G., Silveri, M.C., & Laudanna, A. (1985). Reading mechanisms and the organisation of the lexicon: Evidence from acquired dyslexia. Cognitive Neuropsychology, 2, 81–114.
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.
Carr, T.H., Brown, J.S., & Charalambous, A. (1989). Repetition and reading: Perceptual encoding mechanisms are very abstractive but not very interactive. Journal of Experimental Psychology: Learning, Memory and Cognition, 15, 763–778.
Coltheart, M. (1981). Disorders of reading and their implications for models of normal reading. Visible Language, 15, 245–286.
Coltheart, M. & Funnell, E. (1987). Reading and writing: One lexicon or two? In D.A. Allport, D.G. Mackay, W. Prinz, & E. Scheerer (Eds), Language perception and production:Relationships between listening, speaking, reading and writing. London: Academic Press.
Costello, A. de L. & Warrington, E.K. (1987). The dissociation of visuospatial neglect and neglect dyslexia. Journal of Neurology, Neurosurgery and Psychiatry, 50, 1110–1116.
Diller, L. & Weinberg, J.P. (1977). Hemi-inattention in rehabilitation: The evolution of a rational remediation program. In E.A. Weinstein & R.P. Friedland (Eds), Hemi-inattention and hemisphere specialisation. Advances in Neurology Vol. 18. New York: Raven Press.
Ellis, A.W. (1982). Spelling and writing (and reading and speaking). In A.W. Ellis (Ed.), Normality and pathology in cognitive functions. London: Academic Press.
Ellis, A.W. (1988). Normal writing processes and peripheral acquired dysgraphias. Language and Cognitive Processes, 3, 99–127.
Ellis, A.W., Flude, B.M., & Young, A.W. (1987a). “Neglect dyslexia” and the early visual processing of letters in words and nonwords. Cognitive Neuropsychology, 4, 439–464.
Ellis, A.W., Young, A.W., & Flude, B.M. (1987b). “Afferent dysgraphia” in a patient and in normal subjects. Cognitive Neuropsychology, 4, 465–486.
Friedland, R.P. & Weinstein, E.A. (1977). Hemi-inattention and hemisphere specialisation: Introduction and historical review. In E.A. Weinstein & R.P. Friedland (Eds), Hemi-inattention and hemisphere specialisation. Advances in Neurology Vol. 18. New York: Raven Press.
Gilliat, R.W. & Pratt R.T.C. (1952). Disorders of perception and performance in a case of right-sided cerebral thrombosis. Journal of Neurology, Neurosurgery and Psychiatry, 15, 264–271.
Hillis, A.E. & Caramazza, A. (1989). The graphemic buffer and attentional mechanisms. Brain and Language, 36, 208–235.
Hillis, A.E. & Caramazza. (1991). The effects of attentional deficits on reading and spelling. In A. Caramazza (Ed.), Cognitive neuropsychology and neurolinguistics: Advances in models of language processing and impairments. Hillsdale, NJ: Lawrence Erlbaum Associates Inc.
Kartsounis, L.D. & Warrington, E.K. (1989). Unilateral neglect overcome by cues implicit in stimulus displays. Journal of Neurology, Neurosurgery and Psychiatry, 52, 1253–1259.
Katz, R.B. & Sevush, S. (1989). Positional dyslexia. Brain and Language, 37, 266–289.
Kinsbourne, M. & Warrington, E.K. (1962). A variety of reading disability associated with right-hemisphere lesions. Journal of Neurology, Neurosurgery and Psychiatry, 25, 339–344.
Laplane, D. & Degos, J.D. (1983). Motor neglect. Journal of Neurology, Neurosurgery and Psychiatry, 46, 152–158.
Lebrun, Y. (1976). Neurolinguistic models of language and speech. In H. Whitaker & H.A. Whitaker (Eds), Studies in neurolinguistics, Vol. 1. New York: Academic Press.
Monsell, S. (1987). On the relation between lexical input and output pathways for speech. In D.A. Allport, D. MacKay, W. Prinz, & E. Scheerer (Eds), Language perception and production: Relationships between listening, speaking, reading and writing. London: Academic Press.
Mozer, M.C. & Behrmann, M. (1990). On the interaction of selective attention and lexical knowledge: A connectionist account of neglect dyslexia. Journal of Cognitive Neuroscience, 2, 96–123.
Patterson, K. & Coltheart, V. (1987). Phonological processes in reading: A tutorial review. In M. Coltheart (Ed.), Attention and performance XII: The psychology of reading. Hove: Lawrence Erlbaum Associates Ltd.
Patterson, K. & Wilson, B. (1991). A ROSE is a ROSE or a NOSE: A deficit in initial letter identification. Cognitive Neuropsychology, 7, 447–477.
Quintan, PT. (1991). Connectionism and psychology: A psychological perspective on new connectionist research. Hemel Hempstead: Harvester Wheatsheaf.
Rayner, K., McConkie, G.W., & Zola, D. (1980). Integrating information across fixations. Cognitive Psychology, 12, 206–226.
Riddoch, M.J. (1991). Neglect and the peripheral dyslexias. Cognitive Neuropsychology, 7, 369–389.
Riddoch, M.J. & Humphreys, G.W. (1983). The effect of cueing on unilateral neglect. Neuropsychologia, 21, 589–599.
Riddoch, M.J., Humphreys, G.W., Cleton, P., & Fery, P. (1991). Interaction of attentional and lexical processes in neglect dyslexia. Cognitive Neuropsychology, 7, 479–517.
Warrington, E.K. (1991). Right neglect dyslexia: A single case study. Cognitive Neuropsychology 8,, 193–212.
Warrington, E.K. & Zangwill, O.L. (1957). A study of dyslexia. Journal of Neurology, Neurosurgery and Psychiatry, 20, 208–215.
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. London: Academic Press.
Young, A.W., Newcombe, F., & Ellis, A.W. (1991). Different impairments contribute to neglect dyslexia. Cognitive Neuropsychology, 8, 177–191.