Zambezia (2001), XXVIII (ii).SELECTIVE IMPAIRMENT IN FIRST LANGUAGE WITHPRESERVED SECOND LANGUAGE ARTICULATION: ACASE STUDY*PATRICK CH1RORO, ISAAC MUKURA AND CLARA SHANADepartment of Psychology, University of ZimbabweAbstractIn this article, we report the case of patient KK who, following cerebralinfection, appears to have lost communicative ability in her first languagebut can communicate effectively in a second language. In experiments 1 and2, KKmade a disproportionately greater number of speech production eirorsin response to orthographically and phonologically presented Shona wordscompared to control subjects. No difference was observed between KK'sperformance and that of the control subjects when the same tasks wereadministered using common English words. The results obtained inexperiment 3 showed that although KK found it extremely difficult to readaloud (or repeat after the experimenter) common Shona words, her abilityto access the correct meanings of these words was not impaired. KK'sperformance in a task which required her to select the correct meaning inEnglish of 60 common Shona words was well above chance (90%+ correct).It is argued that KK's problems occur at the phonological output level. Thetheoretical implications of these results are discussed.INTRODUCTIONIt is generally believed that the human language processing systemconsists of relatively independent components which can be selectivelyimpaired by brain damage. For example, studies of monolingual patientswith brain damage have shown differential impairments in the processingof high versus low frequency words (Patterson, Marshall and Coltheart,1985), words of different grammatical classes (Coltheart, Patterson andMarshall, 1980), and words belonging to different semantic categories(Hart, Berndt and Caramazza, 1985; Warrington and McCarthy, 1987).Selective impairments in comprehension and naming restricted to animalsand vegetables have also been reported (see Hillis and Caramazza, 1991).All communication relating to this article should be sent to: Dr Patrick Chiroro, Centre forApplied Psychology, Department of Psychology, University of Zimbabwe, P. O. Box MP167, Mount Pleasant, Harare. Electronic mail should be addressed to: <pchiron>®e$anelzu<*.This research was partially funded by a University of Zimbabwe Research Board Grantawarded to the first author: Grant No. 3YYH10/3787.263264 SELECTIVE IMPAIRMENTHowever, the case of a long-term permanent and all-encompassing selectiveimpairment in articulation of one's first language with preservedarticulation of a second language has, to our knowledge, not been reportedin the literature. In this article, we report the case of patient KK who,following cerebral infection, appears to have lost communicative abilityin her first language but can communicate effectively in a second language.Paradis (1980)' documented cases of bilingual patients whoexperienced a temporary and alternating loss of speech in one languagewith preserved speech in another. One such case involved a 48 year-oldfemale who spoke both French (the first language) and Arabic (the secondlanguage) fluently before she suffered a cerebral concussion in a mopedaccident in Casablanca. Upon regaining consciousness, she was able tospeak in Arabic but, 10 days later, she regained her French but could notspeak in Arabic. On the following day, she again regained her Arabic buther French was very poor while on the 12th day, she was again able tospeak French but not Arabic. The patient was discharged from hospitaland for several weeks, she remained fluent only in French but could notfind her words in Arabic. After three months, the patient recovered herability to speak in both languages.The second patient described by Paradis (1980) was a 23 year-oldmale who spoke both French (the first language) and English (the secondlanguage) fluently before a venous malformation deep in the left parietallobe was surgically removed. For the first 7 days after the operation, thepatient could speak in English but could not speak in French. Later, heregained his French but could no longer speak in English. Subsequently,the patient recovered his ability to speak both French and English.Segalowitz (1983) interprets these findings as strong evidence for separaterepresentations in the brain for different languages. However, it is notclear at present what form these 'different representations' might take.Furthermore, unlike the case described in this article, the studies citedabove showed temporary and alternating recovery of speech in bilingualpatients after which both languages were subsequently restored.Current models of language processing suggest that the lexical systemconsists of modality-specific input and output components that areinterconnected by a general semantic system (e.g. Hillis and Caramazza,1991). Such models assume that whether one reads the word 'car'(orthographic input), sees a car (structural/visual input), or hears theword 'car' (phonological input), the same meaning in the semantic systemis accessed. However, the production of spoken language and writtenlanguage are thought to each involve different components of the lexicalCited in Segalowitz (1983).P. CHIRORO, I. MUKURA AND C. SHANA 265system (see also Ellis, 1992). The dissociation between spoken languageand written language has been demonstrated in several studies. Forexample, it has been shown that in some patients, the ability to namepictures can be impaired yet the same patients can write down names ofthe same pictures with little or no difficulty (Heir and Mohr, 1977; Ellis,Miller and Sin, 1983; and Caramazza and Hillis, 1990). Also, it has beenshown that when the semantic component is impaired, some patientshave difficulty in both spoken and written language (Hillis, Rapp, Romaniand Caramazza, 1990).Such findings suggest, therefore, speech failures can either resultfrom impaired semantic processing or from a failure to access the correctphonological or orthographic representations of known words. In thecase of impaired semantic processing, all forms of language productionwould be affected. Presumably, in bilinguals, both languages would beimpaired. However, if the impairment is due to a failure to access thecorrect phonological representations of words in a particular language,subjects could still be able to access the meaning of phonologically andorthographically presented words yet fail to articulate them orally. In thepresent study, we examined the nature, extent and source of KK's lexicalproblem.CASE DESCRIPTIONKK is a 45 year-old female police officer who, until 1984, spoke bothEnglish (her second language) and Shona (her first language) fluentlybefore she suffered from acute cerebral malaria. In Zimbabwe, Englishserves both as the language of instruction in schools and as the officiallanguage in business, government, and law. Like many Zimbabweans, KKbegan to learn English at the age of 7 years in school. Shona is a localBantu language spoken by over 80% of the indigenous population inZimbabwe. KK was married between 1966 and 1973. However, althoughEnglish is the official language used at KK's work place, both Shona andNdebele are widely used when officers deal with clients who cannotspeak English. She is now divorced and lives with one of her threedaughters and two grandchildren. There is no history of language problemsin KK's family. An examination of KK's medical records showed that soonafter being hospitalised, KK was unable to speak for three days. On thefourth day, she spontaneously regained her speech but could onlycommunicate in English. Since then (more than 10 years ago), KK hasfailed to regain the ability to communicate effectively in Shona, her firstlanguage, this despite very supportive workmates and family members.KK has not received and is currently not receiving any medical orpsychological help regarding her language problem.266 SELECTIVE IMPAIRMENTExperiment 1In this experiment, we compared KK's ability to read aloud commonEnglish and Shona words to that of four control subjects in order toestablish the extent of KK's impairment. On the basis of our informalobservation, we predicted that KK would make a substantially greaternumber of errors on the Shona words than the control subjects but showa comparable level of performance to that of the controls on the Englishwords. The four control subjects who participated in all the experimentsreported in this article matched KK's educational level, profession, maritalstatus, sex, and age. All the control subjects were native Shona speakers.MethodStimuli. First, a list of 60 common English words was prepared. Of thesewords, 20 referred to objects (e.g. broom), 20 referred to animals (e.g.lion), and 20 referred to actions (e.g. walk). A further 20 common Englishnames (e.g. Peter) were added to this list. A "Shona expert" from theDepartment of Linguistics at the University of Zimbabwe was asked tosupply Shona equivalents for each of the 20 object words (e.g. mutsoairo=> broom), 20 animal words (e.g. shumba => lion), and 20 action words(e.g. famba => walk). The same expert also provided 20 common Shonanames (e.g. Tatenda). Thus, altogether, a total of 160 words were used inthe experiment.Apparatus. A NECTAR (486DX-66) IBM-compatible PC was used to presentthe stimuli to KK and to all the control subjects. A programme written inBorland C++ was used to present stimuli on the computer screen. Acomputer was used to present the words in this experiment in order tofamiliarise the subjects with computer-controlled experiments aspreparation for their participation in experiment 3 which required theuse of a computer.Procedure. The subjects were tested individually in a room that wassound-proof at the University of Zimbabwe. Each subject sat next to theexperimenter facing the computer screen on which the words were to bepresented. Subjects were given the following instructions:1 will show you some common English and Shona words on this computerscreen. The words will be shown to you one word at a time at thecentre of the screen. I would like you to read aloud each of the words.When the subject was ready, the experimenter started the programmewhich presented the words. Immediately after each subject's response,the experimenter pressed a button on the keyboard and the next wordappeared. This procedure was followed until all the words had beenpresented. The programme presented the words in a semi-random orderwith the only constraint being that no more than three words from theP. CHIRORO, I. MUKURA AND C. SHANA 267same referral category or language were to appear consecutively.Responses made by each subject were noted by the experimenter interms of whether the response was 'correct' or 'incorrect'.Results and DiscussionThe number of errors made by KK and the mean number of errors madeby the four control subjects in this experiment are shown inTable 1.Table 1NUMBER OF ERRORS MADE BY KK AND MEAN NUMBER OF ERRORSMADE BY FOUR CONTROL SUBJECTS IN EXPERIMENT 1. PERCENTAGEERROR RATES ARE SHOWN IN PARENTHESESREFERRAL CATEGORYActions Objects Animals NamesEnglish WordsKK 2.0(10%) 3.0(15%) 3.0(15%) 0(0%)Controls 3.0(15%) 2.0(10%) 2.0(10%) 0(0%)s.d. 1.13 1.64 ' 2.11 ŠShona WordsKK 11 (55%) 12 (60%) 13 (65%) 0 (%)Controls 1.0(5%) 1.0(5%) 2.0(10%) 0(0%)s.d. 1.54 1.33 2.17 ŠIt can be seen from Table 1 (column 5) that both KK and the controlsubjects made no errors when reading aloud people's names, regardlessof whether these were English names or Shona names. Furthermore, nosignificant difference was observed between KK and the four controlsubjects in their ability to read aloud English words referring to actions,objects, and animals. However, as predicted, KK made a disproportionatelygreater number of errors when reading aloud Shona words referring toactions (55% vs. 5% for controls), objects (60% vs. 5% for controls), andanimals (65% vs. 10% for controls).Three hypotheses may account for KK's errors. First, it could beargued that KK experienced difficulty at the visio-perceptual level interms of encoding the orthographic structure of the letters that make upShona words. Second, it could be argued that KK had difficulty in matchingthe orthographic structure of the Shona words to appropriate phonologicalrepresentations of the familiar words used in this experiment.Contemporary theories of word production distinguish between theavailability of a semantic specification of a word and subsequent retrieval268 SELECTIVE IMPAIRMENTof its phonological form (see Howard, 1995 for an alternative view).Finally, it is possible that KK's impairment is characterised by a failure toproduce the appropriate sounds for the words. In the next experiment,the 'visio-perceptual encoding deficit' hypothesis was tested. Instead ofpresenting written words (orthographic input) to KK and the controlsubjects, a different mode of presentation was used, namely, phonologicalinput.Experiment 2It was hypothesised in this experiment that if KK's Shona articulationproblems are even partially due to failure to encode the orthographicstructure of Shona words, then, presenting the stimuli auditorily mightimprove her performance. However, if KK's Shona articulation problemsemanate from either a failure at matching lexical input to appropriatephonological representations of Shona words or to a failure at thephonological output stage, then, KK should once again show adisproportionate number of errors compared to control subjects in atask in which she is asked to repeat after the experimenter some commonShona words.A second aim of the present study was to establish more preciselythe nature of KK's Shona articulation problems. In the previous experiment,the errors made by KK were simply counted for each referral categoryand for each language. No attempt was made to categorise the errors. Inthe present experiment, KK's errors were categorised into errors involving(p wrong pronunciation (ii) hesitation and halting, (ii) phonemic omission,(iv) phonemic inclusion, and (v) complete failure to say the word. Thesecategories were based on an analysis of KK's responses in the previousexperiment.MethodStimuli and Apparatus. A list of 40 common Shona words was obtainedfrom another "Shona expert" in the Department of Linguistics at theUniversity of Zimbabwe. Of these, 20 were action words and 20 wereproper nouns. None of the words used in the previous experiment wereused m this experiment. The words were printed on white cards in boldblack ink.Procedure. KK and all the four control subjects were tested individuallyin the same room that was used during experiment 1. The experimentersat directly in front of each subject. Between the subject and theexperimenter was a large square table on which the word cards wereplaced in a pile. The following instructions were given to each subject inEnglish:P. CHIRORO, I. MUKURA AND C. SHANA 269With me here I have forty cards. On each card, a word is written. I willpick one card at a time and read to you what is written on the card. 1would like you to simply repeat the word after me until we havefinished all the forty words.The experimenter then read out each of the 20 action words and eachof the 20 proper nouns in random order. As each subject repeated thewords, errors were noted by both the experimenter and an assistant forwrong pronunciation, hesitation and halting, phonemic omission,phonemic inclusion, and failure to say the word.Results and DiscussionTable 2 shows the number of errors made by KK and the mean number oferrors made by the four control subjects in this experiment.Table 2NUMBER OF CORRECT RESPONSES AND ERRORS MADE BY KK ANDMEAN NUMBER OF CORRECT RESPONSES AND MEAN NUMBER OFERRORS MADE BY THE FOUR CONTROL SUBJECTS IN EXPERIMENT 2.PERCENTAGE ERRORS ARE SHOWN IN PARENTHESESAction Words Proper NounsCorrect Failure Others Correct Failure OthersKK 4.0(20%) 9.0(45%) 7.0(35%) 2.0(10%) 11(55%) 7(35%)Controls 20 (100%) Š Š 20 (100%) ŠOnce again, KK made a disproportionately greater number of errorscompared to the control subjects. She was able to say out correctly only20% of the action words and 10% of the proper nouns, compared to 100%accuracy for the controls on both action words and on proper nouns. Ofthe errors she made on action words, 45% involved a 'complete failure' torepeat the target word. Of the errors she made on proper nouns, morethan half (55%) also involved a 'complete failure' to say the target word.The category 'Others' in Table 2 involved such errors as incompletearticulation (e.g. SIM for Simukai), hesitations, and mispronunciations.These results clearly show that KK has a major Shona articulation problem.This problem occurs regardless of whether KK is asked to read aloudShona words (orthographic input - experiment 1) or to repeat Shonawords after someone (phonological input - experiment 2). In both cases,KK made a substantially greater number of errors compared to controlsubjects. Thus, the perceptual encoding deficit hypothesis is notsupported by the results obtained in this experiment.One question that has not been addressed so far is whether KK canattach the correct meaning to a Shona word that is presented270 SELECTIVE IMPAIRMENTorthographically. Failure to attach meaning to Shona words may preventKK from accessing the correct phonological representations for eachword that is presented to her. Therefore, before the next experiment wasconducted, KK was simply asked: "When people talk in Shona, can youunderstand what they will be saying?". KK did not hesitate to provide theanswer "yes" but she qualified her answer by saying "... provided thespeaker is not too fast". This suggested that KK's problem may have littleto do with semantic processing of the Shona input (orthographic orphonological). The next experiment was designed to establish whetherthis is indeed the case.Experiment 3The results obtained in experiments 1 and 2 showed that KK's Shonaarticulation problem occurred regardless of whether the words werepresented orthographically or phonologically. in the present experiment,it was hypothesised that KK's Shona articulation problem may be due toa failure to attach the correct meaning to orthographically andphonologically presented Shona words which, in turn, would prevent herfrom accessing the correct phonological representations of the words. Ifthis hypothesis is to be supported, KK should make a significantly greaternumber of errors than control subjects in a task that requires selectingthe correct English translation for a presented Shona word regardless ofwhether or not KK can say the word aloud. However, if KK is capable ofaccessing correct semantic representations of Shona words, no significantdifference between her performance and that of controls should beobserved in this task.Stimuli and Apparatus. Sixty translated pairs of common English-to-Shona words were used in the experiment. Twenty of the words referredto objects, 20 referred to actions, and 20 referred to animals. These werenot the same words that were used in experiment 1. The presentexperiment was conducted four weeks after the first experiment. A furtherlist of 60 English words were added to this list to serve as distracters. A486DX-66 IBM-compatible PC controlled by a Borland C+ + programmewas used to present the words to the subjects. The programme waswritten in such a way as to randomise the presentation of the words interms of referral category.Procedure. Subjects were tested individually in a specially preparedroom in the Department of Psychology at the University of Zimbabwe.Each subject was asked to sit in front of the computer screen and giventhe following instructions:I am going to show you some common Shona words, one word at a timeon this computer screen. Each word will be shown for 5 seconds. Afterthe word has disappeared from the screen, two English words willP. CHIRORO, I. MUKURA AND C. SHANA 271appear on the screen. I would like you to decide which of the twoEnglish words has the same meaning as the Shona word that you willhave just seen.The experimenter sat next to each subject throughout the experiment.When the subject was ready, the experimenter pressed a key to begin theexperiment. Each subject responded by pointing at the word of her choiceafter which the experimenter entered either "1" for correct responses or"0" for incorrect responses. The computer logged these responses andthe programme calculated the total number of errors made by eachsubject for each of the four referral categories of words.RESULTS AND DISCUSSIONThe results obtained in this experiment are shown in Table 3. For thecontrol subjects, mean and standard deviation scores are given.Table. 3NUMBER OF CORRECT RESPONSES MADE BY KK AND FOUR CONTROLSUBJECTS IN EXPERIMENT 3. PERCENTAGE ERRORS ARE SHOWN INPARENTHESESObjects Actions Animalsn=20 n=20 n=20KK 18.00(90%) 19.00(95%) 19.00(95%)Controls 18.75(94%) 17.75(89%) 19.25(96%)sd 2.14 2.10 1.23It is clear from Table 3 that KK's ability to associate Shona wordswith their correct meanings in English was comparable to that of thecontrol subjects. Also, given the fact that the target Shona words appearedon the screen for only 3 seconds, KK's performance on this task is quiteremarkable. KK made only 2 errors (an error rate of 10%) on the 'objects'list, 1 error (an error rate of 5%) on the 'actions' list, and 1 error (an errorrate of 5%) on the 'animals' list.It can be concluded from these results that KK appears to have nodifficulty what-so-ever in encoding and accessing the meanings oforthographically presented Shona words regardless of the category towhich the words belonged.GENERAL DISCUSSIONThe results obtained in the three experiments presented in this articlestrongly suggest that KK's speech problems occur at the phonological272 SELECTIVE IMPAIRMENToutput level of the lexical system. In experiment 1, KK made a substantiallygreater number of errors compared to control subjects in a task whichrequired subjects to read aloud orthographically presented Shona words.However, KK's ability to read aloud orthographically presented Englishwords was comparable to that of the control subjects. Both KK and thecontrol subjects did not make any errors when they were asked to readaloud common Shona and English names. In experiment 2, KK also madea substantially greater number of errors compared to control subjectswhen asked to repeat after the experimenter 20 action words and 20common nouns in Shona. KK showed no impairment in a similar task inwhich 20 common action words and 20 common nouns were presented inEnglish.Taken together, the results obtained in experiments 1 and 2 suggestthat KK's articulation problem occurs only when she attempts to read orpronounce Shona words and not when English words are involved. Thisis the puzzling bit. It is not possible, on the basis of the results obtainedin experiments 1 and 2, to advance a firm theoretical explanation of KK'serrors. However, the possibility that KK's articulation problem might bedue to failure to encode the meanings of orthographically presentedShona words which, in turn, would prevent access to correct phonologicalrepresentations of the words can be ruled out on the basis of the resultsobtained in experiment 3. In the latter experiment, KK demonstrated anextra-ordinary ability to choose the correct English words for each of the60 Shona words that were presented to her. This, she managed to dodespite the fact that the target Shona words were each presented for only3 seconds. Thus, it could be argued that KK's ability to access themeaning of written Shona words is still intact. Furthermore, KK appearsto have no serious processing deficits at the visual and auditory analysisstages of lexical processing (see Ellis, 1992).In conclusion, the experiments presented here demonstrate twothings. First, the results demonstrate the existence of selective impairmentat the phonological output stage of the lexical system with preservedunderstanding of meanings of words. An opposite effect, that involving aremarkable ability to read words aloud without understanding theirTQJT^I bee" reP°rted jn the literature (Schwartz, Marin, and Saffran,19/9; Schwartz, Saffran and Marin, 1980). Thus, the results obtained inthe present study are consistent with the view that the phonologicalword production system is functionally independent from the semanticsystem. Second, KK's problem, which is characterised by a long-termpermanent selective impairment in first language with preserved secondcSrS tar,tlCulation is surprisingly unique. This type of impairmentcould be taken as evidence that different phonological output systemsare used when communicating in phonetically different languages (a viewP. CHIRORO, I. MUKURA AND C. SHANA 273favoured by Segalowitz, 1983). However, before this theoretical positioncan be accepted, more evidence must be obtained. Use of brain-scanningtechniques might provide such evidence.ReferencesCARAMAZZA, A. AND HILLIS, A. E. (1990) "Where do semantic errors comefrom?", Cortex, 26, 95-122.COLTHEART, M., PATTERSON, K. AND MARSHALL, J. C. (1980) Deep Dyslexia(London, Routledge and Kegan Paul).ELLIS, A. W. (1992) Reading, Writing and Dyslexia: A Cognitive Analysis(London, Lawrence Erlbaum Associates).ELLIS, A. W., MILLER, D. AND SIN, G. (1983) "Wenerkie's aphasia and normallanguage processing: A case study in cognitive neuropsychology",Cognition, 15, 111-144.HART, J., BERNDT, R., AND CARAMAZZA, A. (1985) "Category-specific namingdeficit following cerebral infection", Nature, 301, 338.HEIR, D. B. AND MOHR, J. P. (1977) "Incongruous oral and written naming",Brain and Language, 4, 115-126.HILLS, A. E. AND CARAMAZZA, A. (1991) "Mechanisms for accessing lexicalrepresentations for output: Evidence from a category-specific semanticdeficit", Brain and Language, 40, 106-144.HILLIS, A.E., RAPP, B. C, ROMANI, C. AND CARAMAZZA, A. (1990) "Selectiveimpairment of semantics in lexical processing", CognitiveNeuropsychology, 7, 191-244.HOWARD, D. (1995) "Lexical Anomia: Or the case of missing Lexical entries",The Quarterly Journal of Experimental Psychology: Section A - HumanExperimental Psychology, 48A (4), 999- 1023.PATTERSON, K.E., MARSHALL, J. AND COLTHEART, M. (1985) Surface Dyslexia(London, Lawrence Erlbaum Associates).SEGALOWITZ, J. S. (1983) Two Sides of the. Brain: Brain Lateralization Explored(New Jersey, Prentice Hall), 153-156.SHALLICE, T. AND MCCARTHY, R. A. (1985) "Phonological reading: Frompatterns of performance to possible procedures", in K. E. Patterson,J. C. Marshall and M. Coltheart (eds.) Surface Dyslexia (London,Lawrence Erlbaum Associates).SCHWARTZ, M. F. MARIN, O.S.M., AND SAFFRAN, E.M. (1979) "Dissociations oflanguage function in dementia: A case study", Brain and Language, 7,277-306.SCHWARTZ, M. F., SAFFRAN, AND E. M. MARIN, O.S.M. (1980) "Fractionatingthe reading processes in dementia: Evidence for word-specific print-to-sound associations", in M. Coltheart, K.E. Patterson and J. C.Marshall (eds.) Deep Dyslexia (London, Routledge and Kegan Paul).WARRINGTON, E. K. AND MCCARTHY, R. A. (1983) "Category-specific accessdysphasia", Brain, 106, 859-879.Š (1987) "Categories of knowledge: Further fractionations and anattempted explanation", Brain, 110, 1273-1296.