Phonological alexia: three dissociationsalexia beingdistinguished (BensonandGeschwind, (1969): alexia with agraphia, aphasic alexia, and pure alexia (or alexia without either agraphia

Journal ofNeurology, Neurosurgery, and Psychiatry, 1979, 42, 1115-1124

Phonological alexia: three dissociationsM. F. BEAUVOIS AND J. DEROUESNE

From Laboratoire de Neuropsychologie INSERM U.84, Hopital de la Salpetriere, Paris, France

SUMMARY Three dissociations were observed in a case of alexia: a disturbance of reading,without comparable disturbance of oral expression, oral comprehension, writing, or spellingaloud; a disturbance of the phonological reading process, without disturbance of the non-phonological reading process; a disturbance located at the level of the phonological stage, withoutdisturbance of the perceptual and expressive stages. This pattern of results has been calledphonological alexia.

Most research on alexia has until recently beencarried out from a syndromic and anatomoclinicalpoint of view. This has led to a broadly acceptedclassification of alexia in terms of the clinicalsymptoms which are usually found in conjunctionwith it. This classification conforms in essenceto that of Dejerine (1891) with three varieties ofalexia being distinguished (Benson and Geschwind,(1969): alexia with agraphia, aphasic alexia, andpure alexia (or alexia without either agraphia oraphasia). Each variety of alexia is ascribed to adistinct neuroanatomical locus of lesion in the leftcerebral hemisphere-alexia with agraphia isascribed to a parietal lesion, aphasic alexia to atemporo-parietal one, and pure alexia to a lesion ofthe left occipital lobe, associated by many authorswith a lesion of the splenium of the corpus callo-sum. However, there are problems with thisclassification. Firstly, recently reported cases ofalexia (Albert et al., 1973; Benson, 1977) cannotbe included in any one of the three varieties.Secondly, any one variety can be interpreted inseveral ways. For example, alexia without agraphiais thought of either as consecutive to a defect invisual perception (Alajouanine et al., 1960;Lhermitte and Gautier, 1969) or to a visuoverbaldisconnection (Geschwind, 1965). Thus, thisclassification can neither account for numerousreading impairments which result from cerebrallesions, nor permit distinction of reading disturb-ances caused by different dysfunctions in any onevariety of alexia. For about 10 years, some studieson alexia have been carried out from a linguistic

Address for correspondence and reprint requests: Dr M. F. Beauvois,Laboratoire de Neuropsychologie INSERM U.84, H6pital de la Sal-petri6re, 47 Bd de l'H6pital, 75634 Paris Cedex 13, France.Accepted 12 May 1979

or a psycholinguistic point of view. Three tendenciescan be distinguished among them. Some investiga-tors have attempted to find facts confirming orinvalidating particular aspects of a general theoryof linguistic competence (Marshall and New-combe, 1966). Others, on the basis of a methodof linguistic analysis, have tried to describe read-ing disorders more systematically. For example,following Hecaen (1967) in applying the principlesof distributional analysis, Dubois-Charlier (1971,1972) and Kremin (1976) proposed a classificationof alexia according to linguistic level-they dis-tinguished literal alexia, verbal alexia, and phrasticalexia. Finally, others have considered alexia as ameans of studying the production of reading be-haviour itself in terms of the information process-ing systems involved (Marshall and Newcombe,1973; Shallice and Warrington, 1975; Pattersonand Marcel, 1977; Saffran and Marin, 1977). Theresearch reported here has been carried out fromthis last point of view.

In this report we use a model of the readingprocesses derived from an organigram proposedby Andreewsky (1974) and related to certainhypotheses on reading behaviour suggested duringthe previous few years (Albert et al., 1973;Marshall and Newcombe, 1973; Shallice and War-rington, 1975; Patterson and Marcel, 1977;Schwartz et al., 1977; Saffran and Marin, 1977).In the model, reading is assumed to involve atleast two processes which must be distinguished atthe perceptuovisual level as well as at the linguisticlevel.The first reading process is analytical at the

visual level and phonological at the linguistic level.The proposed sequence is as follows: first, thereis an analytical, letter-by-letter or syllable-by-

1115

guest. Protected by copyright.

on January 17, 2021 byhttp://jnnp.bm

j.com/

J Neurol N

eurosurg Psychiatry: first published as 10.1136/jnnp.42.12.1115 on 1 D

ecember 1979. D

ownloaded from

M. F. Beauvois and J. Derouesne

syllable visual processing; then a linguistic cor-respondence between these visually identifiedletters and explicitly or implicitly uttered soundsis effected; the word can be understood only whenthis analytical processing is effected for all oralmost all of its letters. Thus, at the linguistic levelsuccess in this reading process depends on a phono-logical stage (grapheme-phoneme correspondence)which precedes and conditions lexical and semanticstages. This is why this process is called phono-logical reading process.On the other hand, the second reading process

is global at the visual level, and lexical or semanticat the linguistic level. The proposed sequence isas follows: at the visual level the different lettersof a word are understood simultaneously andglobally, and are recognised as a whole (or asubset of letters is understood and constitutes aclue which is sufficient to evoke the word); then alinguistic correspondence between this identifiedunit and a particular word is effected. At thelinguistic level, success in this reading processdepends first on a lexical or semantic stage whichprecedes and conditions the phonological stage, ieimplicit verbalisation, so that the verbalisation ofa word does not necessarily precede its compre-hension; for Andreewski (1974) it is even probablethat comprehension precedes verbalisation. Con-sequently, not all written stimuli can be read bymeans of this reading process; only lexical andmeaningful ones can. This process is called lexicalor non-phonological reading process.

In the last few years several neuropsychologicalstudies bearing on cases of a reading impairmentwhich has been called either deep (Marshall andNewcombe, 1973) or phonemic (Shallice and War-rington, 1975) dyslexia have contained argumentsfor the existence of two reading processes. Twotypes of research have been done. Firstly, the des-cription of this type of alexia made it possible topostulate the existence of a non-phonologicalreading route. Experimental research then estab-lished that patients with deep dyslexia did manifestimpairment of the phonological reading route(Patterson and Marcel, 1977; Saffran and Marin,1977), which showed that their performances werecertainly linked to the existence of a non-phonological route. At the present time someproblems remain. Firstly, these researches wereall carried out on aphasic patients. Consequently,these patients' errors were not necessarily relatedto the supposed impairment. For example, thepresence of semantic paraphasias in reading aloudmay be caused by an impairment of stages locatedafter the establishment of the grapheme-phoneme

correspondence (related to aphasia) and not bya patient's inability to establish this correspon-dence. Secondly, the level at which impairment ofthe phonological route is situated has not alwaysbeen specified experimentally. Thirdly, the possibleeffects of the disturbance upon other linguistictasks have never been studied.

In this paper we report a neuropsychologicalstudy of a case of alexia without expressive orreceptive dysphasia, which was carried out on thebasis of the reading behaviour model previouslydescribed. The experiment was conducted toanswer the following questions. 1. Can the twopostulated reading processes work independently?In order to answer this, the question was askedwhether serious but incomplete alexia could bedue to a single reading process disturbance. 2.Was the reading process disturbed at the level ofvisual perception, verbalisation, or at the inter-mediate linguistic level? 3. Were other activities,which might have been assumed to involve thestage disturbed, normally performed by thepatient?

Case summary

As the detailed case report and its anatomical datahave been reported by Beauvois et al. (1978), onlya short summary is given here.The patient (RG) was a right handed, 64 year

old man, a retired agricultural machinery repre-sentative. Two years before the neuropsychologicalstudy, he had been operated upon for a leftparieto-occipital angioma. Computerised tom-ography scans showed a lesion involving the leftangular gyrus, the posterior part of the secondtemporal convolution, the inferior longitudinalfasciculus, the geniculostriate fibres, and somefibres of the tapetum. At the time of this study,besides the alexia, he was still suffering from thefollowing neurological defects: right inferiorquadrantanopia, mild memory deficit, mild cal-culation impairment, mimimal constructionalapraxia, and astereognosia. On the other hand, hesuffered from neither motor nor sensory defect;his oral expression and comprehension could beconsidered to be good, with the exception of thebilateral tactile aphasia previously reported.The alexia was serious enough to prevent the

patient from resuming his work, and had continuedwithout any change for two years. While he hadbeen a passionate reader, RG could no longerread a paper or a letter he had just written. TheAlouette reading test (Lefavrais, 1963) confirmedhis serious reading disability: his reading level was

1116

guest. Protected by copyright.

on January 17, 2021 byhttp://jnnp.bm

j.com/

J Neurol N

eurosurg Psychiatry: first published as 10.1136/jnnp.42.12.1115 on 1 D

ecember 1979. D

ownloaded from

Phonological alexia

that of a 6 year old child-that is, the level ofthe lowest class at primary school. This was dueto very slow reading (in this text, RG could notread more than 62 stimuli in three minutes) andto the presence of numerous errors (40% of thestimuli were misread or not read at all) whichinvolved particularly articles, possessive adjectives,conjunctions, and some verbs.

In contrast with his very impaired reading, hiswriting was fairly good with the exception ofsome spelling mistakes. He could write down amessage and draw up a text on a particular topic,being impeded only by the fact that he could notread over what he had written.

Neuropsychological study

FUNCTIONAL INDEPENDENCE OF THE TWO READING

PROCESSESFirst procedureEach of the two reading processes was excludedsuccessively at the perceptual level, by experi-mental procedures making impossible eitheranalytical or global word perception.The patient was asked to read aloud two lists

of nouns. In the first, the words were presented inhandwritten form. The letters themselves werehardly recognisable, so that the word could beidentified practically only from its global form. Inthe second, the word letters were very legible(they were capital letters in print form) but theirorder was reversed (for instance, BOUTEILLE waswritten ELLIETUOB) in order to prevent any globalword recognition and to make an analytical per-ception necessary. Words were read aloud nearlyperfectly by means of global perception (19/20);none was read aloud by means of analytical per-ception (0/20), although the patient was givenseveral minutes to do so.

Second procedureThe phonological reading process was handled atthe linguistic level. When the stimuli are meaning-ful, the lexical reading process is sufficient, andthe phonological reading process is unnecessary;but when the stimuli are meaningless, the phono-logical reading process becomes necessary. Thus,the reading of meaningless stimuli-that is, ar-bitrary strings of letters or digits-was comparedwith the reading of meaningful stimuli-that is,nouns or meaningful numbers.Test I Reading aloud of letter strings The sub-ject had to read aloud a list of 40 nouns (each offive to nine letters) and a list of 40 non-words(each of four or five letters-for example, PUKO,

DIRMA). All the nouns were read correctly, com-pared with only 10% of the non-words. Nounswere read quickly whereas non-words were readvery slowly. Misreading of non-words resultedfrom either a failed attempt to decode analyticallyor an attempt to reconstitute the non-word froma word which looked like it (VINA: "c'est presquevinaigre", ie, "it's almost vinegar"; FOTA: "ondirait presque football", ie, "it looks like football").Test 2 Reading aloud of numbers The subjectwas asked to read aloud two lists of 10 four-digitnumbers. In the first, the numbers had no obviousmeaning (for example, 4358); in the second, theywere meaningful (for instance, the year of thepatient's birth, the dates of the second world war).The patient was told that the numbers of the firstlist were meaningless and the numbers of thesecond meaningful. No meaningless number wasread correctly; all meaningful numbers were cor-rectly and quickly read.Test 3 Pointing out a letter string correspondingto an orally presented stimulus The patient wasasked to designate from 20 letter strings, the onecorresponding to a stimulus pronounced by theexperimenter. He had to point out 40 nouns and40 non-words consecutively. All the nouns werepointed out correctly, compared with only 30%of the non-words.

Third procedureThe variables affecting the necessarily phonologi-cal reading of non-words were researched and itwas verified that they did not affect the reading ofwords, which was assumed to be achieved bythe lexical reading process. Two variables-stimulus length and syllabic composition-weremanipulated.Stimulus length The patient was asked to readaloud two lists of stimuli. The first list containednon-words: 40 single letters, 40 two-letter non-words, and 40 four- or five-letter non-words. Thesecond list contained nouns including 20 shortones (three letters) and 20 long ones (six to nineletters). The longer the non-words, the worse theywere read (Table 1). There was no significant dif-ference between reading performance for shortand long nouns.Syllabic composition The order of vowels andconsonants in a word could affect reading diffi-culty. Stimuli of simple syllabic composition-thatis, in which each consonant is followed by onevowel (CVCV) might be easier to read thanstimuli in which consonants were found in suc-cession (CCVC). So, the effect of syllabic com-position was studied in reading performance for

1117

guest. Protected by copyright.

on January 17, 2021 byhttp://jnnp.bm

j.com/

J Neurol N

eurosurg Psychiatry: first published as 10.1136/jnnp.42.12.1115 on 1 D

ecember 1979. D

ownloaded from

M. F. Beauvois and J. Derouesne'

Table 1 Effect of stimulus length on readingperformance. Percentage of correct responses

Type of word Stimulus length

Non-word One letter Two letters Four or five letters87% 42% 10%

Noun Three letters Six to nine letters85% 100%

words (nouns and adjectives) and non-words. Thetest on words was composed of two lists of stimuli.In the first list (20 words), the syllabic compositionwas assumed to be simple (for example, domino,securite). In the second one (40 words) the syllabiccomposition was assumed to be complex (forexample, absolu, frustre). The two lists wereequivalent as regards certain characteristics oftheir words- length, frequency, grammaticalcategory. The test on non-words was composed oftwo lists of 40 stimuli. In the first, the syllabiccomposition was simple (for example, VORA). Inthe second, it was complex (for example, STO).The non-words having a simple syllabic com-position were made up of four letters; the non-words having a complex composition were madeup of three letters. Syllabic composition signifi-cantly affected reading of non-words. It did notaffect reading of words (Table 2).

It is clear, both from the model proposed above,and from the results obtained in patients withdeep dyslexia, that not all categories of wordscan be read aloud by means of the lexical readingprocess. From the model, the fact that wordscan or cannot be read by the global process woulddepend on their meaning (the words having littlemeaning would not be read), and the results ob-tained in patients with deep dyslexia are affectedby variables such as frequency, concreteness, andparts of speech. Since the aim of this study wasto show the opposition between the phonologicaland the non-phonological reading processes, andnot to test the efficiency of the global process, allthe words chosen for the tests described abovewere high frequency (20 per million; Gougenheim

Table 2 Effect of syllabic composition on readingperformance. Percentage of correct responses

Simple Complex Comparisoncomposition composition X2 test

Non-words eg: VORA eg: STO X2=5.54,P <.0555% 32%

Words eg: FACILrrTE eg: FRUSTRE NS75% 75%

S=significant; NS=not significant.

et al., 1964), concrete nouns (with the exceptionof the test studying the effect of syllabic compo-sition), which were highly meaningful. It is impor-tant to underline the fact that this patient couldnot read all the single words, but only about 80%of them. The variables affecting his word reading,which were different from those affecting readingby patients with deep dyslexia will be presented indetail elsewhere.These three kinds of experiment did establish

the functional independence of the two readingprocesses in this patient. Indeed, on the one handthere was a disturbance of the phonological read-ing process without a disturbance of the lexicalreading process. On the other hand, the variablesaffecting the reading of non-words did not affectthe reading of words.

DISTURBANCE AT THE PHONOLOGICAL STAGE AND NOTAT THE PERCEPTUAL OR EXPRESSIVE STAGESIn order to determine whether the disturbancewas actually located at the level of the phono-logical stage allowing the grapheme-phonemecorrespondence and not at the level of the per-ceptual or expressive stages, it was necessary toestablish two points: (1) that the analytical visualprocessing was performed-in order to test thiswe verified that the reading disturbance was notspecific to the visual presentation, that the lettercould be visually identified, and that it could beindividualised in a word; (2) and that single lettersand non-words could be pronounced in othertasks.

Experimental design on single lettersThe aim of this experiment was to verify that thealexia was related neither to a specific (visual)mode of presentation of the stimulus nor to an(oral) response modality, but to an interactionbetween them both. A systematic design includingsix tests was thought of in order to manipulatetwo modalities of response (oral and visual) andthree modalities of stimulus presentation (visual,gestural, and auditory). Because of this patient'sbilateral tactile aphasia, the tactile presentationwas not studied. In each test 40 letters werepresented consecutively.The three tests in which the response was oral

were as follows: reading aloud of seen letters,reading aloud of letters presented by gesture, andrepetition test. The three tests in which the responsewas visual were pointing out tests. The patientwas asked to point out, on a board comprising26 letters, the one corresponding to the targetletter. The tests were pointing out upon spoken

1118

guest. Protected by copyright.

on January 17, 2021 byhttp://jnnp.bm

j.com/

J Neurol N

eurosurg Psychiatry: first published as 10.1136/jnnp.42.12.1115 on 1 D

ecember 1979. D

ownloaded from

Phonological alexia

request, visuovisual matching, and gesturovisualmatching ("show me the letter which has thesame name as the one I have just drawn withyour hand"). In the latter two tests, the form ofthe stimulus (handwritten small letters in scriptform) was different from the form of the response(handwritten capital letters in print form). Thusthese tests did not study simple perceptual in-formation transmission from one modality toanother.The test in which the stimulus was auditory and

the response was oral (repetition) was carried outperfectly (Table 3), as was the test in which boththe stimulus and the response were visual (visuo-visual matching). The results recorded for all theother tests were impaired.

Control tests on lettersThe transmission of simple perceptual informa-tion from one modality to another was studied. Inorder to do this, it was tested whether the letterA drawn by the experimenter with the patient'shand could be matched with a visually presentedA. The test included 40 matchings of identicalletters. They were all done perfectly. This resultmeans that the possible errors in matching be-tween two modalities in the above experiment(a =A), must not be thought of as a defect ofinformation transmission from one modality toanother one.

Identification of a letter included in a non-wordTwo tests were carried out. In test 1 the patienthad to match a single small letter in script formwith a capital letter in print form included in a10-letter non-word (for example, a single f withF included in LOUFREVANT), and 50/50 correctresponses were recorded. In test 2 the patient hadto decide whether two non-words were identical ornot. If they were not, he had to indicate preciselythe nature of the difference. The non-words werecomposed of six to 10 letters. When two non-

Table 3 Effect of the presentation modality and ofthe response modality on letter reading performance.Percentage of correct responses

StimulusResponse Visual Gestural Auditory

Oral Reading aloud Reading aloud Repetitionseen letters letters drawn with 100%87% the patient's hand

83%Visual Visuovisual Gesturovisual Pointing out

matching matching upon spoken100% 75% request

80%

words were different, their difference came onlyfrom one letter. Either they included a differentletter (for example, ZANROLIER and sanrolier) orthey included the same letters, but the placing ofone of the letters was different (for example,PINFAGUE and pifangue). The test was constructedso that it was not possible to succeed by simpleperceptual matching. For instance the letter whichwas different in two non-words was never i or o.The 20 proposed comparisons were executed per-fectly. Whenever the non-words were different, thepatient was quick to show the difference.

Repetition of 40 non-words (composed of eightletters).All were repeated perfectly.

The results permit rejection of the hypothesisof alexia caused by a visual impairment, for thefollowing reasons: (1) the reading defect was notrestricted to the visual modality (reading aloudwas as impaired with gestural presentation as withvisual presentation); (2) visual identification ofletters was well performed since the patientmatched perfectly two visually different forms ofthe same letter; (3) individualisation of a letter ina non-word was performed perfectly.The results also permit rejection of the hypo-

thesis of alexia caused by a verbalisation im-pairment. On the one hand, the patient mademistakes in tasks requiring no verbalisation, that is,in tests involving pointing out upon spoken request(visuovisual matching excepted). On the otherhand, he could repeat letters and non-words with-out any mistake. The reading impairment occurredwhenever the task brought into play both letterform identification (no matter whether the pres-entation modality was visual or gestural) and theproduction of a sound corresponding to this form,or inversely, the identification of a sound corres-ponding to a letter form and the identification ofthis letter form. Thus, the reading impairmentwas located at the level of the phonological pro-cessing necessary to the grapheme-phonemecorrespondence.

WAS THE DISTURBANCE SPECIFIC TO READING?Study of other kinds of linguistic behaviourOral expression Spontaneous speech was correct.Articulation, intonation, and fluency were good,and sentences were constructed correctly. Thepatient repeated the 50 syllables, the 125 words,and the 10 sentences of the Test pour l'Examende l'Aphasie (Ducarne, 1976) entirely correctly.He could repeat four two-syllable words in thecorrect order. He gave correct word definitions

1119

guest. Protected by copyright.

on January 17, 2021 byhttp://jnnp.bm

j.com/

J Neurol N

eurosurg Psychiatry: first published as 10.1136/jnnp.42.12.1115 on 1 D

ecember 1979. D

ownloaded from

M. F. Beauvois and J. Derouesne

(standard score 9 on WAIS Vocabulary Subtest).He easily constructed sentences using two or threegiven words (score 7/7 on the sentence construc-tion test of the Test pour l'Examen de l'Aphasie).He quickly stated the opposite of a given word49 times out of 50. He was able to name 50 objectimages out of 50 and 39 sounds of objects out of40, but he could correctly name only two objectsout of three when these objects were presentedtactually. With the exception of this bilateraltactile aphasia, which has been previously reported(Beauvois et al., 1978), the patient's verbal ex-pression was normal.Oral language understanding The patient wasable to carry out correctly all the tests of "oralunderstanding" from the Test pour l'Examen del'Aphasie-pointing out an object named by theexaminer (35/35), carrying out instructions (9/10),and the Three Papers Test of Pierre Marie. Thecritique of irrational stories of the TermanMerrill Intelligence Scale was performed cor-rectly up to the highest level. A discriminationtest on closely related phonemes (Derouesne andSaillant, unpublished) was performed correctly.Semantic discrimination was good (IQ: 98 for theSynonym Test of Binois and Pichot (1958)orally performed). Therefore, the patient's oralcomprehension was good.Writing The patient was able to compose andwrite a story, being impeded only by the fact thathe could not read over what he had written. Hecould take down dictated words, sentences, andeven a text. All this was performed correctlyexcept for a moderate dysorthographia. The wordtranscription was always correct phonetically, butsome spelling mistakes were observed (forexample, enfant was written ENFANS, eglise waswritten AIGLISE). Thus, the patient's writing wasnot entirely normal although it was very muchbetter than his reading.

Therefore, not all linguistic tasks were carriedout perfectly by this patient since he had abilateral tactile aphasia and a moderate dysortho-graphia. However, on the one hand thesedisturbances were very specific and on the otherhand they appeared to have no direct relationwith his reading impairment. With the exceptionof the language impairment observed when objectswere tactually presented, his oral expression andcomprehension were completely normal, and it ishard to conceive what relation except a contiguousone could exist between tactile aphasia and alexia.The writing disturbance consisted only of a milddysorthographia. This excepted, writing was com-pletely normal, and in particular the phonetictranscription which involves phonological process-

ing was normal in all the clinical tests. However,some verifications were done in order to establishthat phonological processing was actually carriedout in other different tasks bearing on writtenlanguage.

Experimental study of phonological processinginvolved in tasks bearing on written languageThis was studied by means of the following tests.Writing from dictation The patient had to writefrom dictation 40 non-words composed of sixletters. All were written perfectly.Copying The patient had to copy 40 non-wordscomposed of four or five letters. Stimuli werewritten in print form letters, and the patient hadto write them in script form. They were copiedvery badly (22% correct responses), and most ofthe errors were similar to those made in reading.Pronunciation of words and non-words orallyspelled out by the experimenter The experi-menter spelled out a string of letters and askedthe subject to pronounce the non-word corres-ponding to that string. The test included 40 non-words composed of four or five letters.Performance was good enough but not perfect(83% correct responses). The minimal difficultyin carrying out this test was only due to the factthat the patient had difficulty in keeping in mindfive or more letters. Indeed, all the four-letter non-words were pronounced perfectly, while with longnon-words either one letter was forgotten or theend of the non-word was wrongly guessed.Spelling aloud The patient had to spell aloud theletters corresponding to a stimulus pronouncedby the experimenter. The test included 40 non-words composed of six letters. All were spelledcorrectly.The only test which was wrongly carried out

was the one involving reading (copying); all othertests were done correctly. Therefore phonologicalprocessing was carried out in all the tasks bearingon written language which did not involvereading.

Discussion

The results recorded for this patient can be sum-marised into three essential dissociations: (1) adisturbance of the phonological reading process,without disturbance of the lexical reading process;(2) a disturbance located at the level of the phono-logical stage that permits grapheme-phoneme cor-respondence, without disturbance at the perceptualand expressive stages (3) a disturbance of readingwithout comparable disturbance of oral expression,oral comprehension, writing, and spelling aloud.

1120

guest. Protected by copyright.

on January 17, 2021 byhttp://jnnp.bm

j.com/

J Neurol N

eurosurg Psychiatry: first published as 10.1136/jnnp.42.12.1115 on 1 D

ecember 1979. D

ownloaded from

Phonological alexia

Such a pattern of results, which we have chosento call phonological alexia, has important impli-cations for the understanding of normal behaviouras well as of reading impairments resulting fromcerebral lesions.

READING IN NORMAL SUBJECTS

Independence of the two reading processes

The chief result of this study concerns the experi-mental validation of the possible independentfunctioning of two reading processes. Our patientwas found to be suffering from a disturbance ofthe phonological reading process, but without dis-turbance of the lexical reading process. This dis-turbance was related to a dysfunction of thephonological stage which allows grapheme-phoneme correspondence. Thus, these results con-

firm data concerning a non-phonological route inreading, obtained in deep dyslexia (Marshall andNewcombe, 1973), also called phonemic dyslexia(Shallice and Warrington, 1975; Patterson andMarcel, 1977; Saffran and Marin, 1977). Thesedata mean that firstly, the lexical reading process

does exist and is likely to function independentlyfrom the phonological reading process, andsecondly, that this independent functioning in-volves the intermediate linguistic stages and notonly the visual or expressive stages.However, one of our results differs from data

obtained by these authors. Words which were used(nouns and adjectives) were read nearly perfectly,so that the often reported semantic errors were

not observed. In our view, this is because our

patient had only a very specific language disturb-ance which has been called tactile aphasia, buthad no expressive or receptive aphasia, contraryto the other reported cases. Indeed, if an alexicpatient is also aphasic, presumably some of hislexical disturbances (for example, difficulty withword evocation, semantic paraphasia) will impedethe functioning of the lexical reading process. Inthis case, it was not possible to determine if theword misreading was due to the dysfunctioning ofthe lexical reading process, or to the fact that thisprocess in itself did not allow the subject to readwords perfectly. Our results provide evidence insupport of the idea that the lexical reading processis sufficient to read at least the words used inthis study. Thus, the problem of how our patientcan read words, and what sort of words he can

read, is an important matter in establishing thedegree of efficiency of the global reading process.Independence of speech compared with readingBilateral tactile aphasia excepted, the patient'sspeech was completely normal. This means thatthe phonological processing necessary to the

1121

grapheme-phoneme correspondence in reading iscompletely different from that necessary to oralexpression and comprehension.

Independence of writing from dictation comparedwith readingNon-words which were not read aloud werewritten perfectly from dictation. The discrepancybetween normal writing and severely impairedreading has often been described since it charac-terises what is called alexia without agraphia.In this kind of alexia, the disturbance is thoughtto be located at the visual level or at the levelof visuoverbal transmission, so that it is easy toimagine that there is no agraphia since visualperception is not very important in writing. Thus,it could be imagined that writing and reading weresymmetrical activities involving the same stagesbut in reverse order, and differing only by theirend points-visual perception and graphicrealisation. In particular one could assume thatwriting and reading involve the same linguisticstages. This concept is illustrated by some ex-planations of alexia with agraphia. Hecaen (1967),for example, thought that the reading impairmentwas caused by the same disturbance as the writ-ing impairment, that is, a loss of grapheme-phoneme correspondence. Our experiments showedthat a loss of grapheme-phoneme correspondencemay be found in reading and not in writing. Con-sequently, writing and reading can no longer beconceived as symmetrical activities. This em-phasises that a written grapheme is not equivalentto a seen grapheme. It is not merely more diffi-cult to produce a grapheme from dictation thanto identify it visually. These activities are com-pletely different, in particular at the linguisticlevel.

Relationship between copying and readingConfronted with letters and non-words, the patientcopied correctly when he was asked to write astimulus in the same written form as that in whichthe stimulus was presented (copying withoutretranscription). He copied with many mistakeswhen he was asked to change capital letters inprint form into small letters in script form. Thiscontrasts, on the one hand, with good writingfrom dictation and on the other hand with theretained ability to match two visually presenteddifferent forms of the same letter or of the samenon-word. It is possible that the copying impair-ment was related to another disturbance than thatwhich provoked the alexia. However, it is hardto imagine what phonological stage could be in-volved in copying and not in writing from dicta-

guest. Protected by copyright.

on January 17, 2021 byhttp://jnnp.bm

j.com/

J Neurol N

eurosurg Psychiatry: first published as 10.1136/jnnp.42.12.1115 on 1 D

ecember 1979. D

ownloaded from

1122

tion and in reading. It is more probable that thecopying impairment was related directly to thedisturbance provoking the phonological reading im-pairment. This should mean that a normal subjectcarries out the correspondence between a seenletter and a drawn letter of different form bymeans of phonological processing. Gestural pro-duction of a grapheme from a different visuallypresented form of this grapheme requires first theevocation of the corresponding phoneme.Independence of spelling aloud compared withreading The patient, who could not read non-words, could spell them perfectly aloud if theywere orally presented. This means that spellingaloud does not necessarily involve the evocationof the "visual image" of the stimulus whoseletters would be then read one after the other.Indeed, the patient had some difficulty in readingletters, but none in spelling aloud non-wordspresented orally. Thus, in order to spell aloud aword or a non-word, other strategies than theevocation of the "visual" image are efficient; forinstance the subject can implicitly pronounce (orhear) the stimulus and cut it up into phonemes,which will be translated into letter names.

READING DISTURBANCES (ALEXIA)Phonological alexia differs from the differentforms of alexia that have already been described.

Firstly it differs from each of the three usualsyndromic or anatomoclinical varieties of alexia.It cannot be included under aphasic alexia, since,with the exception of a bilateral tactile aphasia,this patient had no language impairment. It ishard to imagine what relationship (except a con-tiguous one) could exist between tactile aphasiaand alexia. Besides, the patient showed no signusually found in conjunction with aphasic alexia(Alajouanine et al., 1960; Benson and Geschwind,1969; Lhermitte and Gautier, 1969). Some tasksgenerally more disturbed than reading tasks inalexic aphasic patients were, in the present case,done well enough-spontaneous writing, writingfrom dictation, spelling aloud, and the pronunci-ation of non-words orally spelled out by the ex-perimenter. The problem of comparison withparietal alexia (also termed alexia with agraphia)may be raised because of the locus of the lesion(angular gyrus) which was the one usually re-ported in patients suffering from this sort ofalexia. But there is an important difference. Ourpatient had no agraphia, but only a milddysorthographia, and, contrary to what has oftenbeen described in alexia with agraphia, his writingdisturbances did not resemble his reading disturb-ance. Whereas in reading, he could not carry

M. F. Beauvois and J. Derouesne

out the grapheme-phoneme correspondence neces-sary to the phonological reading process, he coulddo so perfectly in writing. Words were always wellwritten phonetically, and he could write down non-words from dictation. Finally, this patient's readingimpairment differs from pure alexia. In purealexia, letters and syllables are usually better readthan words, whereas the reverse pattern was ob-served here; in pure alexia, words are read ana-lytically, letter by letter or syllable by syllable(Alajouanine et al., 1960; Lhermitte and Gautie,1969). Our patient proceeded in a reverse way bytrying to guess non-words from words whichlooked like them. Besides, the interpretation ofthis disturbance does not fit the usual interpreta-tions of pure alexia. It was clearly established thatthe disorder reported here cannot be thought of asconsecutive to a visual processing defect as postu-lated in pure alexia (Lhermitte and Gautier, 1969),since every strictly visual test was carried out per-fectly. Neither can the alexia reported here beconsidered as consecutive to a visuoverbal discon-nection (Benson and Geschwind, 1969) in thecommon understanding of this term. This dis-connection could not be general since writtenwords could be read aloud. It should concernonly analytically and phonologically processedinformation. This is theoretically possible sincetwo independent reading processes have beenpostulated, but does not correspond to what isgenerally implied by the term visuoverbal dis-connection. Besides, the locus of the lesion doesnot fit that of pure alexia. The computedtomography scans (Beauvois et al., 1978) showeda left hemispheric parietal lesion, and there wasno evidence of occipital or callosal lesion. Inaddition, the patient only had a quadrantanopia, sothat visual information reached the left occipitalarea

Secondly, phonological alexia differs from thetwo forms of alexia which have been distinguishedaccording to the linguistic level of disturbance-verbal alexia and literal alexia (Wernicke, 1974;Hecaen, 1967; Dubois-Charlier, 1971, 1972;Kremin, 1976). The reading impairment for non-words which characterises phonological alexia hassometimes been included in verbal alexia (Dubois-Charlier, 1971). This was because non-words werethought to be located at the same linguistic levelas words. Obviously, for us, phonological alexia,as described here is completely different fromverbal alexia since words were read almost per-fectly. But more often, phonological alexia hasbeen confused with literal alexia, probably becausethe impaired reading of non-words or syllableswas then conceived as consecutive to impairment

guest. Protected by copyright.

on January 17, 2021 byhttp://jnnp.bm

j.com/

J Neurol N

eurosurg Psychiatry: first published as 10.1136/jnnp.42.12.1115 on 1 D

ecember 1979. D

ownloaded from

Phonological alexia

of letter reading. In fact, in the case reportedhere, when letters had to be read one by one,they were read well enough. It was when theyhad to be integrated into syllables, and thus had aphonological value, that they could not be read.This emphasises an important differentiation instudies of alexia. There are two possible strategiesin reading letters, one phonological, the otherlexical. Each of these strategies is adapted to aparticular kind of stimulus. When the subjecthas to read a single letter, the most efficientstrategy is the lexical one, since he is asked toname it. When he has to read a letter includedin syllables or non-words, the most efficientstrategy is the phonological one, since he has tomake grapheme-phoneme correspondence. Thus,the term literal alexia is confusing.

Finally, from a descriptive point of view,phonological alexia differs from deep (Marshalland Newcombe, 1973) or phonemic dyslexia(Shallice and Warrington, 1975). Indeed, for someauthors "the key to the demonstration of phonemicdyslexia is the appearance of semantic errors"(Schwartz et al., 1977). Yet, our patient did notmake any semantic errors. On the other hand, theinterpretation of phonemic dyslexia is similar tothe interpretation of phonological alexia since itshould be caused by an impairment of the "pho-nemic" route, the semantic route being spared. Ifthis is the case, the term phonological seems to usbetter than the term phonemic. In the studiescarried out up to now, the impairment has beenlocated at the level of the phonological stage allow-ing grapheme-phoneme correspondence withoutthere being anything in the experiments whichpermits us to decide if it is on the phonemic siderather than on the graphemic side. The distinctionbetween two kinds of phonological impairment inreading (the graphemic and the phonemic) is dis-cussed by Derouesne and Beauvois (1979).But in spite of this similarity, there is an im-

portant difference between phonemic dyslexia andphonological alexia. The first is a clinical entity,the other a theoretical concept. Phonemic dyslexia(Shallice and Warrington, 1975) was originally de-fined by means of a list of errors (semantic, deri-vational and visual ones) and of stimuli sup-posed to be misread (for example, abstract words,function words) which have no direct relationwith the location of the disturbance on the readingmodel. They are reported only because they haveoften been found in association with the phono-logical impairment. By contrast, phonologicalalexia is defined only in relation to the readingmodel (one reading process is disturbed, the other

1123

is not). Phonological alexia has been revealedhere in a spectacular manner thanks to a rarepatient who could be considered as a "pure case".In most clinical cases, phonological alexia isprobably associated with other disturbances. Inparticular, it is probable that the association ofphonological alexia with lexical alexia (ie, impair-ment of word reading) gives the clinical picturewhich has been called phonemic dyslexia.

We thank J. K. Hylton from the Universite deParis XII for help in translation and Dr T.Shallice for his helpful criticisms.

References

Alajouanine, T., Lhermitte, F., and de Ribeaucourt-Ducarne, B. (1960). Les alexies agnosiques etaphasiques. In Les grandes activites du lobeoccipital, pp. 235-265. Edited by T. Alajouanine.Masson: Paris.

Albert, M. L., Yamadori, A., Gardner, H., and Howes,D. (1973). Comprehension in alexia. Brain, 96, 317-328.

Andreewsky, E. (1974). Un modele semantique.Application a la pathologie du language: alexiaaphasique. Traitenzent A utomatique. Informations,2, 3-27.

Beauvois, M. F., Saillant, B., Meininger, V., andLhermitte, F. (1978). Bilateral tactile aphasia: atacto-verbal dysfunction. Brain, 101, 381-401.

Benson, F. (1977). The third alexia. Archives ofNeurology (Chicago), 34, 327-331.

Benson, D. F., and Geschwind, N. (1969). The alexias.In Handbook of Clinical Neurology, vol. 4, pp. 112-140. Edited by P. J. Vinken and G. W. Bruyn.North-Holland Publishing Company: Amsterdam.

Binois, R., and Pichot, P. (1958). Test de Vocabulaire.Editions du Centre de Psychologie Appliquee:Paris.

Dejerine, H. (1891). Contribution a 1'etude anatomo-pathologique et clinique des differentes variet6s dececite verbale. Memoires de la Societe de Biologie,4, 61-90.

Derouesne, J., and Beauvois, M. F. (1979). Phono-logical processing in reading. Data from alexia.Journal of Neurology, Neurosurgery, andPsychiatry, 42, 1125-1 132.

Dubois-Charlier, F. (1971). Approche neurolinguis-tique du probleme de l'alexie pure. Journal dePsychologie Normale et Pathologique, 1, 39-68.

Dubois-Charlier, F. (1972). A propos de l'alexie pure.Languages, 25, 76-94.

Ducarne de Ribeaucourt, B. (1976). Test pourl'examen de l'aphasie. Editions du Centre dePsychologie Appliquee: Paris.

Geschwind, N. (1965). Disconnexion syndromes inanimals and man. Brain, 88, 237-294.

Gougenheim, G., Michea, R., Rivenc, P., and

E

guest. Protected by copyright.

on January 17, 2021 byhttp://jnnp.bm

j.com/

J Neurol N

eurosurg Psychiatry: first published as 10.1136/jnnp.42.12.1115 on 1 D

ecember 1979. D

ownloaded from

M. F. Beauvois and J. De'rouesne

Sauvageot, A. (1964). L'elaboration du fran,caisfondamental. Didier: Paris.

Hecaen, H. (1967). Aspects des troubles de la lecture(alexies) au cours des lesions cerebrales en foyer.Word, 23, 265-287.

Kremin, H. (1976). Les problemes de l'alexie pure.Langages, 44, 82-110.

Lefavrais, P. (1963). Test de l'alouette. Editions duCentre de Psychologie Appliquee: Paris.

Lhermitte, F., and Gautier, J. C. (1969). Aphasia. InHandbook of Clinical Neurology, vol. 4, pp. 84-104.Edited by P. J. Vinken and G. W. Bruyn. North-Holland Publishing Company: Amsterdam.

Marshall, J. C., and Newcombe, F. (1966). Syntacticand semantic errors in paralexia. Neuropsychologia,4, 169-176.

Marshall, J. C., and Newcombe, F. (1973). Patternof paralexia: a psycholinguistic approach. Journalof Psychlolinguistic Research, 2, 175-200.

Patterson, K. E., and Marcel, A. J. (1977). Aphasia,dyslexia and the phonological coding of writtenwords. Quarterly Journal of ExperimentalPsychology, 29, 307-318.

Saffran, E. M., and Marin, 0. S. M. (1977). Readingwithout phonology: evidence from aphasia.Quarterly Journal of Experimental Psychology, 29,515-525.

Shallice, T., and Warrington, E. K. (1975). Wordrecognition in a phonemic dyslexic patient.Quarterly Journal of Experimental Psychology, 27,187-199.

Schwartz, M. F., Saffran, E. A., and Marin, 0. S. M.(1977). An analysis of agrammatic reading inaphasia. Paper presented at the Meeting of theInternational Neuropsychological Society. Santa Fe,February 1977.

Wernicke, K. (1874). Der aphasische Symptomen-complex. Franck and Weigert: Breslau.

1124

guest. Protected by copyright.

on January 17, 2021 byhttp://jnnp.bm

j.com/

J Neurol N

eurosurg Psychiatry: first published as 10.1136/jnnp.42.12.1115 on 1 D

ecember 1979. D

ownloaded from