Quarterly Journal of Experimental Psychology

ISSN: 0033-555X (Print) (Online) Journal homepage: http://www.tandfonline.com/loi/pqje19

Phonemic dyslexia: Errors of meaning and the meaning of errors K. E. Patterson To cite this article: K. E. Patterson (1978) Phonemic dyslexia: Errors of meaning and the meaning of errors, Quarterly Journal of Experimental Psychology, 30:4, 587-607, DOI: 10.1080/14640747808400688 To link to this article: http://dx.doi.org/10.1080/14640747808400688

Published online: 29 May 2007.

Submit your article to this journal

Article views: 33

View related articles

Citing articles: 74 View citing articles

Full Terms & Conditions of access and use can be found at http://www.tandfonline.com/action/journalInformation?journalCode=pqje19 Download by: [Universite Laval]

Date: 07 November 2015, At: 17:46

Quarterly Journal of Experimental Psychology (1978) 30, $37-601

PHONEMIC DYSLEXIA: ERRORS OF MEANING AND THE MEANING OF ERRORS

Downloaded by [Universite Laval] at 17:46 07 November 2015

K. E. PATTERSON MRC Applied Psychology Unit, 15 Chaucer Road, Cambridge CB2 zEF, U.K. Most paralexic errors made by phonemic dyslexic patients in reading single words aloud are classifiable as derivational (murriuge 3 “married”), semantic (projector -+ “camera”), or visual (pivot + “pilot”) errors. A study of two such patients’ assessment of their own reading showed differences as a function of error type, with the patients generally identifying semantic paralexias as errors but classifying derivational and visual paralexias as correct readings. This pattern was confirmed in a test of the patients’ ability to select one of three spoken words to match a printed word. The results are discussed in terms of a theoretical account of the processing of written words by normal readers and by phonemic dyslexics.

Introduction Deep dyslexia (Marshall and Newcombe, 1973) or phonemic dyslexia (Shallice and Warrington, I 975) is an acquired reading disorder, resulting from brain damage, which has been described by the above-mentioned authors and several others as well (e.g. Patterson and Marcel, 1977; Saffran and Marin, 1977). Some of the notable features of this syndrome include effects, on success of reading words or sentences aloud, of part of speech (nouns > adjectives > verbs > function words) and word imageability (high imageability > low, with frequency controlled). Also the ability to “sound out” written words is impaired, as indicated by an almost complete failure to pronounce orthographically regular non-words like dake given visual presentation. This pattern of reading impairment is naturally of interest to those who study the pathology of language; because the pattern has a number of features seemingly germane to an understanding of the normal reading process, it is also receiving attention from investigators concerned with modelling this process (e.g. Allport, 1977; Coltheart, 1977; Marcel and Patterson, 1978). One of the most intriguing features of deep dyslexia is the high frequency of single-word paralexic errors, where the patient is asked to read a single word aloud and responds with a different word. Paralexic errors represent a situation in which, with no time pressure, with no context, and with the target word in full view, a previously fluent reader simply produces the wrong word. Most phonemic dyslexic patients make at least three apparently distinct types of paralexic error: (I) semantic errors (e.g. conscience-+ “honesty”) with a semantic relationship but no physical similarity between target and response ; (2)visual errors (e.g. badge + 0033-555~/78/040587

+

21

$OZ.OO/O

@ 1978 The Experimental Psychology Society

Downloaded by [Universite Laval] at 17:46 07 November 2015

588

K. E. PAlTERSON

“bandage”) with a visual or graphemic similarity between the two words; and (3) derivational errors (e.g. speak-, “speech”), where the response word has the same root morpheme as the target word but a different derivational form. On an intuitive basis these error types appear to be very different kinds of events. Both semantic and derivational errors seem to preserve the approximate semantics of the target word, for example, while visual paralexias such as quarrel-+ “squirrel” do not. And the different error types are probably attributable to the failure of different components in the process of word recognition and production. Yet there has in fact been little detailed analysis of such errors, little attempt to explore situations where the various types of paralexia might “behave” differently and thus reveal something about their meaning. The purpose of this paper is to explore some facets of the meaning of paralexic errors. The error analysis presented here is based in part on data regarding a patient’s awareness of his own reading errors. Most of the publications on deep dyslexia report, anecdotally, that sometimes a patient realizes (and sometimes even indicates) that he has produced an incorrect reading but at other times seems confident that his erroneous response is correct. In a study of two deep dyslexics reported below, knowledge of correctness of reading responses was assessed by confidence ratings, and knowledge of word identity was assessed by forced choice recognition. In both techniques, the analysis will focus on performance as a function of type of paralexic error, semantic, visual or derivational.

Description of patients The patients who provided the data reported here are two severely aphasic males who come to the Speech Therapy Department of Addenbrooke’s Hospital, Cambridge, on a regular out-patient basis. Clinical and psychological details of these patients have been published previously (Patterson and Marcel, 1977) and only a brief summary will be given here. D.E. (born 1954) had a motor scooter accident in 1970, producing virtually complete occlusion of the left internal carotid artery. P.W. (born 1908)suffered a major left-hemisphere cerebrovascularaccident in 1965. Both patients display right-sided hemiplegias ; neither shows a visual field defect. The Boston Diagnostic Aphasia Test (Goodglass and Kaplan, 1972) yields profiles reasonably characteristic of Broca’s aphasia for both patients. Spontaneous speech is agrammatic ; auditory comprehension, while measurably impaired, is well preserved relative to speech production for both D.E. and P.W. For each patient, a corpus of single-word paralexic errors has been collected from performance on a large number of different word lists. Some of these lists come from standard reading tests such as the Schonell(1942) graded word reading test; other lists derive from standardized tests of language function like the Peabody Picture Word Vocabulary Test (Durn, 1965); still other lists were created for various purposes such as assessing the influence of rated imageability on success of reading (see Patterson and Marcel, 1977). The lists contained nouns, verbs and adjectives, covering a wide range of frequencies of usage, and totalling around 800 words. Over the entire set of words, D.E.’s reading performance was as follows : correct-48% ; paralexic err0r-q.X ;omission (generally “no” or “don’t

589

PHONEMIC DYSLEXIA

know”)-q% ; circumlocution or other kind of response (e.g. an unintelligible word)-I%. For P.W., the corresponding figures are: correct-p% ; paralexic error-37% ; omission-8% ; circumlocution and other-s%. TABLE I D.E.’s single-word paralexias, collected from various reading tests (nouns,verbs, adjectives)

Downloaded by [Universite Laval] at 17:46 07 November 2015

Number Total Semantic Visual Derivational Vis and/or Sem Vis then Sem ?

Proportion

Example

I2

0.23 0.35 0.32 0.06

3

0.01

5

0.025

thermos -+ “flask” idiot +“idol” excavate -+ “excavations” opponent 3 “opposite” favour 3 “taste” decay -+ “ruler”

7.03 47

72 64

A description of the corpus of paralexic responses is shown in Table I for D.E. and in Table I1 for P.W. As mentioned in the introduction, there are three main categories of error (semantic, visual, and derivational), and the tables provide numbers, proportions, and examples of these types. There are several further types of error listed in the tables. “Vis and/or Sem” refers to cases where the response word is both visually and semantically similar to the target word. “Vis then Sem” refers to responses which seem to arise from a visual error followed by a semantic error, e.g. favour+ (flavour)+ “taste” and brought --f (bought)+ “buying”. Finally, a few errors foil classification, requring a “7” category. TABLE I1 P. W.,s single-word paralexias, collected from various reading tests (nouns, verbs, adiectives)

Total Semantic Visual Derivational Vis and/or Sem Vis then Sem ?

Number

Proportion

294 I 60 38

0’54 0.13

64

0‘22

I3

0‘04

I

18

0.06

Example

negative -+ “minus” hassock + “hammock” -+ “editor” edition -+ “segment” fragment brought -+ “buying” encumbered -+ “broken”

While these latter three groups are not without interest (see Shallice and McGill, 1978,for a discussion of “Vis and/or Sem” errors), the three main types account for 90% of the paralexic errors of both patients and are the only types with sufficient frequency to warrant representation in the analyses to follow. With respect to criteria for classification: (I) any cases where the response and target word appeared to have their root morpheme in common were considered

Downloaded by [Universite Laval] at 17:46 07 November 2015

590

K. E. PATTERSON

derivational errors; (2) visual errors were usually obvious, but in cases of any doubt were based on the criterion that at least half of the letters in the response word must be present in the target word; (3) semantic errors defy objective rule-making, but a fairly strict criterion was applied. Also an independent judge was asked to classify the errors, and the few instances of disagreement between judge and author were shifted to the “I”category. Viewed from a perspective of the broad range of language and reading dysfunctions, D.E. and P.W. present highly similar cases, but they also differ in various ways. (I) As indicated in the general summary of their reading behaviour given above, the two patients differ on the relative proportion of omission and commission errors: almost all of P.W.’s failures to read a word correctly are paralexias, while D .E.’s reading yields paralexic errors and omissions in roughly equal proportion. (2) As indicated in Tables I and 11, D.E.’s paralexias consist of a large number of visual and derivational errors, with a smaller though still notable number of semantic errors; in contrast, more than half of P.W.’s paralexias are semantic, with a fair number of derivational errors and relatively few visual ones. (3) Perhaps related to (I), wherever response criterion is a component of the task (e.g. in yes-no as opposed to forced-choice paradigms), D.E. applies a more conservative criterion than P.W. D.E. tends, in other words, to be a “nay-sayer”. (4) Although both patients were previously described (Patterson and Marcel, 1977) as showing essentially normal word/non-word discrimination, more extensive testing on lexical decision (Patterson, in press) indicates that only P.W.’s performance justifies that description. If words are abstract and of low frequency, D.E. fails to recognize many of them as words. Some of these differences between the two patients may turn out to be conceptually significant, and will be reintroduced at subsequent points in the paper.

Experiment I

Method Over several test sessions, the patients were asked to read lists of single words using a confidence rating procedure. The instructions were as follows: “When you read words aloud, as you know, your reading is often exactly correct but sometimes not quite correct. Sometimes you say a different word from the one I’ve asked you to read. I would like to find out, when you read, how certain you are that the word you’ve produced is the word you’re looking at. So after you try to read each word in this list, I will ask you to tell me either that you’re Sure that you’ve said the right word, or you think maybe your response is correct, or you feel the word you’ve said is probably not the same as the word you’re looking at”. Neither patient had any difficulty with the procedure. Typically they did not volunteer the confidence rating information, and so after each reading response the experimenter asked “are you sure?”. Both patients (and P.W. especially) were reluctant to use the middle rating category. D.E. would occasionally respond “think so” or “half-and-half” to the question “are you sure?”, and these responses were assigned to the middle category. But for the most part, they answered the question either “yes” (or “sure”) or “no”. The total number of words included in the confidence rating sessions was about 350 for D.E., 300 for P.W. Only correct readings and single-word paralexic errors are included in the analysis, since confidence ratings are inappropriate for omissions and circumlocutions.

591

PHONEMIC DYSLEXIA

Results and discussion Table I11 presents the number of correct and paralexic responses, and the proportions of them receiving each of the three confidence levels. Both patients were usually confident about their correct readings, though it is certainly of interest to note that they occasionally responded “maybe” or even “no” after a perfectly correct reading. The patients were considerably less confident about their TABLE I11

Downloaded by [Universite Laval] at 17:46 07 November 2015

Proportion of the three types of conJidence rating given to correct readings and to paralexic errors

N Sure Maybe

No

=

correct (156)

D.E. paralexia

correct

P.W. paralexia

(105)

(153)

0.83

0.52

0.94

0.63

0.08 0.08

0.09 0.39

0‘02

0.05

0’04

0-32

(91)

paralexic readings, but even in these cases the most likely rating was “sure”. The data substantiate observational reports by other investigators (Marshall and Newcombe, 1973; Shallice and Warrington, 1975) that phonemic dyslexic patients are sometimes aware, sometimes unaware of reading errors. It is not necessarily the case under ordinary circumstances (i.e. when not asked to give confidence ratings) that patients try to assess the correctness of their reading. Whatever checking procedure they utilize as a basis for confidence ratings, however, often fails to detect a mismatch between printed word and spoken response, and sometimes even fails to detect a match. The analysis of primary interest concerns confidence ratings on paralexic errors as a function of the classification into semantic, visual, and derivational errors (Table IV). For both patients (but more dramatically for P.W.), Table IV suggests that the behaviour of semantic errors is different from that of the other two types. The generalization stated above, that the most likely response to a

TABLE IV

-

Proportion of the three types of confidence rating for different types of paraIexic error ( S = semantic, V = visual, D = derivational) D.E.

N=

S

V

(13)

(54

D (39)

P.W.

r S

V

(42)

(12)

D (34)

Sure Maybe

0.3 I

0‘52

0’59

0.38

0.83

0-15

0.06

0’10

0.05

0.17

0.88 0.03

No

0-54

0-42

0’31

0.57

0’00

0.09

Downloaded by [Universite Laval] at 17:46 07 November 2015

592

K. E. PATTERSON

paralexic error was “sure”, is true for both visual and derivational errors; semantic errors however were most frequently followed by a confidence rating of “no”. For a statistical evaluation of this result, “maybe” and “no” responses were pooled and contrasted to “sure” responses. The probability of a “sure” response given a semantic error as compared to a visual or derivational error was reliably different for P.W., ~ ~ ( 1=) 22-65, P t o - o o ~but , not for D.E., x2(1) = 2.66,0.1o

Phonemic dyslexia: errors of meaning and the meaning of errors.

Quarterly Journal of Experimental Psychology ISSN: 0033-555X (Print) (Online) Journal homepage: http://www.tandfonline.com/loi/pqje19 Phonemic dysle...
2MB Sizes 0 Downloads 0 Views