Full Terms & Conditions of access and use can be found athttp://www.tandfonline.com/action/journalInformation?journalCode=pcgn20

Download by: [Bar-Ilan University] Date: 27 June 2017, At: 02:09

Cognitive Neuropsychology

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

Morpho-syntactic processing of Arabic pluralsafter aphasia: dissecting lexical meaning frommorpho-syntax within word boundaries

Tariq Khwaileh, Richard Body & Ruth Herbert

To cite this article: Tariq Khwaileh, Richard Body & Ruth Herbert (2015) Morpho-syntactic processing of Arabic plurals after aphasia: dissecting lexical meaning frommorpho-syntax within word boundaries, Cognitive Neuropsychology, 32:6, 340-367, DOI:10.1080/02643294.2015.1074893

To link to this article: http://dx.doi.org/10.1080/02643294.2015.1074893

Published online: 05 Oct 2015.

Submit your article to this journal

Article views: 164

View related articles

View Crossmark data

Citing articles: 1 View citing articles

Morpho-syntactic processing of Arabic plurals after aphasia: dissecting lexicalmeaning from morpho-syntax within word boundaries

Tariq Khwaileha*, Richard Bodyb and Ruth Herbertb

aDepartment of English Literature and Linguistics, Qatar University, Doha, Qatar; bDepartment of HumanCommunication Sciences, The University of Sheffield, Sheffield, UK

(Manuscript received 15 October 2013; revised manuscript received 8 February 2015;revised manuscript accepted 16 July 2015)

Within the domain of inflectional morpho-syntax, differential processing of regular and irregularforms has been found in healthy speakers and in aphasia. One view assumes that irregular formsare retrieved as full entities, while regular forms are compiled on-line. An alternative view holdsthat a single mechanism oversees regular and irregular forms. Arabic offers an opportunity tostudy this phenomenon, as Arabic nouns contain a consonantal root, delivering lexicalmeaning, and a vocalic pattern, delivering syntactic information, such as gender and number.The aim of this study is to investigate morpho-syntactic processing of regular (sound) andirregular (broken) Arabic plurals in patients with morpho-syntactic impairment. Threeparticipants with acquired agrammatic aphasia produced plural forms in a picture-naming task.We measured overall response accuracy, then analysed lexical errors and morpho-syntacticerrors, separately. Error analysis revealed different patterns of morpho-syntactic errorsdepending on the type of pluralization (sound vs broken). Omissions formed the vast majorityof errors in sound plurals, while substitution was the only error mechanism that occurred inbroken plurals. The dissociation was statistically significant for retrieval of morpho-syntacticinformation (vocalic pattern) but not for lexical meaning (consonantal root), suggesting thatthe participants’ selective impairment was an effect of the morpho-syntax of plurals. Theseresults suggest that irregular plurals forms are stored, while regular forms are derived. Thecurrent findings support the findings from other languages and provide a new analysistechnique for data from languages with non-concatenative morpho-syntax.

Keyword: aphasia; agrammatism; inflection;morphology; syntax; morpho-syntax; Arabic; regular;irregular; lexical retrieval; plural; dual; sound; broken; non-concatenative; language processing

1. BackgroundRegular and irregular forms are morpho-syntacticstructures that belong to the same grammaticalcategory but differ in terms of their formationprocesses. While regular forms follow one canoni-cal rule in their construction, irregular forms areidiosyncratic. This phenomenon can be found in

both nouns and verbs. For example, Englishregular verb tenses inflect via suffixation (talk;talked), while most irregular forms are formed bychanging the vocalic pattern of the verb (sing;sung) or no change (put; put). English also hasmore complex forms such as “teach, taught” and“bring, brought” (Bybee & Slobin, 1992).

© 2015 Taylor & Francis

*Corresponding author. Email: Tariq.khwaileh@qu.edu.qa

Cognitive Neuropsychology, 2015Vol. 32, No. 6, 340–367, http://dx.doi.org/10.1080/02643294.2015.1074893

This phenomenon has been of interest toresearchers within the domains of inflectionalmorpho-syntax and its breakdown. In languagesother than Arabic, people with aphasia have beenfound to show single and double dissociationsbetween regular and irregular forms in wordproduction (e.g., Bird, Lambon Ralph, Seidenberg,McClelland, & Patterson, 2003; Braber, Patterson,Ellis, & Lambon Ralph, 2005; Lambon Ralph,Braber, McClelland, & Patterson, 2005; Miozzo,2003; Patterson, Lambon Ralph, Hodges, &McClelland, 2001; Penke, Janssen, & Krause,1999; Ullman et al., 1997). Numerous theorieshave been proposed to account for this pattern ofperformance (e.g., Joanisse & Seidenberg, 1999;Pinker & Ullman, 2002; Rumelhart & McCleland,1986; Sereno & Jongman, 1997; Ullman et al.,1997). Theories proposed to account for regularand irregular inflectional processing can be classi-fied into two major frameworks: the dual (e.g.,Pinker & Ullman, 2002; Ullman et al., 1997) andsingle mechanism accounts (e.g., Joanisse & Sei-denberg, 1999; Rumelhart & McCleland, 1986;Sereno & Jongman, 1997). The current studydoes not aim to test all existing models/theoriesbut focuses on the general principles of two viewson regular and irregular processing (Joanisse & Sei-denberg, 1999; Ullman et al., 1997).

1.1. The dual mechanism account

The dual mechanism account has been defined as “afamily of psycholinguistic models which hold thatmorphologically complex word forms can be pro-cessed through stored full-form representationsand by rules that decompose or parse inflected orderived word forms into morphological constitu-ents” (Clahsen, 2006; p. 2). It also postulates thatword forms can be processed associatively(stored) through rule-based (computed) mechan-isms (Clahsen, 1999; 2006; Pinker, 1999; Pinker& Ullman, 2002).

Dual mechanism advocates agree on the pres-ence of these two routes. However, there is dis-agreement concerning the types of word formsprocessed within each route. Chialant and Cara-mazza (1995) suggest that high-frequency wordforms are processed via the stored full-form route,

while the rule-based route (online computation) isresponsible for production of novel or low-fre-quency word forms. Chialant and Caramazza’s(1995) proposal does not differentiate betweenword forms based on their morphological features,but merely on their psycholinguistic features—e.g.,frequency of occurrence. Chialant and Caramazza(1995)’s AAM theory suggests that whole wordform representations may be available for allword forms regardless of frequency. More recently,Cholin, Rapp, and Miozzo (2010) propose thestem-based assembly model (SAM), suggestingthat affixation processes apply more broadly,without limitation to the productive or defaultaffixes. In their proposal, they maintain that combi-natorial affixation is available to all stems, and thatthe lexicon contains an entry for each word formthat includes the basic stem.

An alternative view differentiates betweenprocessing routes on the basis of word form mor-phology (Pinker, 1999; Pinker & Ullman, 2002;Ullman, 2001; Ullman et al., 1997). This viewmaintains that regular inflection is the predictableform, which can be applied to novel word formsand invokes a combinatory rule (e.g., English pasttense regular form rule: stem + -ed; e.g., print-ed).Conversely, irregular inflections are stored formsthat cannot be predicted by the form of the stemand cannot be applied to novel word forms (e.g.,meet; met). This view assumes that the dualmodel uses an associative system that depends ondeclarative memory. It also includes a combinator-ial system that depends on rules to formulate multi-morphemic word forms. It maintains that thelexicon is a subdivision of memory and that irregu-lar forms are stored in memory. However, grammaris a productive and combinatorial system thatcarries out operations that assemble morphemesinto word forms. The associative system general-izes on the basis of resemblance to stored examples,while the combinatorial system suppresses differ-ences between items and treats members of amorpho-syntactic category equally. According tothis model, word forms with regular inflection arecomposed using the combinatorial system inwhich relevant affixes are combined with stems inline with pre-defined rules forming a givenmorpho-logical unit from its basic components (morphemes).

Cognitive Neuropsychology 341

Word forms with irregular morpho-syntax areretrieved from declarative memory. Full-form rep-resentations for morphologically complex wordforms are also retrieved from declarative memory.

1.2. The single mechanism account

The single mechanism account maintains that onemechanism governs the production of bothregular and irregular forms (Bird et al., 2003;Braber et al., 2005; Bybee, 1995; Joanisse & Sei-denberg, 1999; Juola & Plunkett, 2000; LambonRalph et al., 2005; Marchman, 1993; Pattersonet al., 2001; Rumelhart &McClelland, 1986; Stock-all &Marantz, 2006; Yang, 2000). While there are anumber of different views within this family ofmodels, our focus will be on Joanisse and Seiden-berg (1999), who proposed a connectionistaccount based on computerized simulations ofregular and irregular processing. This view isbased on the Parallel Distributed Processing(PDP) model (Rumelhart & McClelland, 1986).The connectionist perspective maintains that bothforms are computed by a single mechanism,which is not rule-based, as this is not necessaryfor the formulation of regular forms. Instead, theprocessing of regular and irregular forms is gov-erned by two types of lexical information: semanticand phonological. According to Joanisse and Sei-denberg (1999), an impairment at producingregular forms results from a phonological deficit,but a deficit in irregular forms results from a seman-tic deficit. They assume that semantic knowledgehas an impact on irregular word forms. They stipu-late that the semantic system retrieves specific idio-syncratic forms of verbs. For example generatingthe past tense of the English verb “take” is an idio-syncratic property of this verb, which requiresspecifically identifying “took” as its past tense.Damage to the semantic representation of thisverb will necessarily result in deficient productionof its past tense, “took”. Therefore phonologicallysimilar word forms, such as “fake; faked”, can beused as analogies that will result in the productionof errors such as “take, *tak-ed”. Phonology hasan impact on regular and novel word forms. Itapplies analogies to word forms that undergosimilar phonological processes (such as adding the

English past-tense verb suffix -ed). For example, anovel word form such as “yug” has no meaning(thus no semantic reference), but its past tense canbe generated using a phonological analogy.A deficit in phonology affects such processes.

Advocates of this account maintain that it is notnecessary to propose differential processing ofregular and irregular forms, even though double dis-sociations between regular and irregular have beenreported (e.g., Miozzo, 2003; Ullman et al., 1997).They maintain that dissociations could be a resultof phonological similarities and differences ratherthan morphological features. They argue that thetwo types of inflectional morpho-syntax—regularand irregular—are not categorically different: theyhave a common structure in two respects. First,there are systematic phonological similaritiesbetween, for example, regular and irregularEnglish verbs. Both present- and past-tense verbslike “take; took” maintain the onset and coda.Second, there are similarities in patterns of the irre-gular past-tense forms “slept” /slɜpt/ and “crept”/krɜpt/. These are similar to the regular past-tenseforms of verbs like “stepped” /stɜpt/ and“cropped” /krɒpt/ (Bird et al., 2003). It is arguedthat the English regular past-tense verbs havesimilar phonological endings, while irregulars donot. Therefore, a phonological effect could resultin dropping the –ed in regular past tense, suggestingthat dissociation between regular and irregular ispresent.

Various studies have investigated differentaspects of the dual and single mechanism frame-works (English: Braber et al., 2005; LambonRalph et al., 2005; Miozzo, 2003; Patterson et al.,2001; German: Penke et al., 1999: Hebrew:Berent, Pinker, & Shimron, 1999; Italian:Walenski,Sosta, Cappa, & Ullman, 2009; Spanish andCatalan: Rodríguez-Fornells, Münte, & Clahsen,2002). Support for both theoretical frameworkshas been found. While findings from these studieshave confirmed certain aspects of a given model,they were inconsistent with other assumptions ofthe same model. For example, deficient irregularinflections in agrammatic aphasia (reported in DeDiego Balaguer, Costa, Sebastián-Gallés, Junca-della, & Caramazza, 2004) indicated that the dualmechanism approach does not account for all

342 T. Khwaileh et al.

patterns of performance. Moreover, irregularimpairment in the presence of phonological deficitand absence of semantic deficit in, for example,AW (Miozzo, 2003) is problematic for Joanisseand Seidenberg’s (1999) single mechanism model.

Spanish and Catalan have regular and irregularverb forms. Furthermore, some regular verbs inSpanish are irregular in Catalan, and vice versa.De Diego Balaguer et al. (2004) investigated theprocessing of regular and irregular verb forms inJM and MP (bilingual speakers of Spanish andCatalan) through a morphological transformationtask in which the participants were asked toprovide appropriate verb forms in sentence frames(e.g., “Ayer yo comía, Hoy yo____” “Yesterday Iate, today I___”). Both JM and MP had Broca’saphasia with agrammatism following cerebrovascu-lar accident. According to the dual mechanismaccount, agrammatic participants should processirregular forms more successfully than regularforms (Pinker & Ullman, 2002; Ullman et al.,1997). The dual mechanism account assumes thatthe breakdown of regular forms is due to impairedrule-based processes linked to a general syntacticprocessing impairment underlying agrammatism.De Deigo Bulaguer et al. (2004) reported that bothJM andMP showed poorer performance of irregularthan regular forms in both Spanish andCatalan. Thisfinding is contradictory to the prediction of the dualmechanism and is consistent with results reported inthe literature (e.g., MR: Laiacona & Caramazza,2004; Penke et al., 1999;RC: Shapiro&Caramazza,2003). De Deigo Bulaguer et al. (2004) concludethat certain aspects of the dual mechanism accountneed to be reconsidered and revised, such as theclaim that agrammatic participants process irregularforms better than regular ones.

Regular and irregular forms are present in theItalian verb system—i.e., past-participle andpresent-tense. Studies have examined theproduction of these two verb aspects in brain-damaged participants (Say & Clahsen, 2002;Walenski et al., 2009). Walenski et al. (2009)explored regular and irregular production inItalian verb tenses (past participle and presenttense) in 12 participants with Alzheimer’sDisease. Participants were asked to read sentencesaloud, filling in the appropriate form of the verb.

All the participants showed impaired irregular pro-duction, while the production of regular forms wasspared. These findings replicate findings fromstudies in English (Miozzo, 2003; Pinker &Ullman, 2002; Ullman et al., 1997). Walenskiet al. (2009) interpreted their findings within thedual mechanism account.

Only one study investigated regular and irregu-lar forms in Arabic after aphasia (Mimouni,Kehayia, & Jarema, 1998). However, the authorsused a theoretical linguistic approach—a prosodicnonconcatenative morphology developed byMcCarthy (1975)—rather than models of proces-sing to account for their data. In their study onAlgerian Arabic, Mimouni et al. (1998) investi-gated the process of word form recognition ofsingular and plural nouns in 24 healthy speakersand two participants with aphasia. The aim was toexamine the lexical representation of plurals inAlgerian Arabic. They used an auditory lexicaldecision task within a morphological priming para-digm. They presented participants with differentsets of word forms in their singular and pluralforms. The stimuli were selected according toplural type (sound vs. broken) and length. Fre-quency values for Arabic are not readily available;therefore they asked healthy participants to rate thefrequency of their stimuli prior to conducting theexperiment. They did not control for any other psy-cholinguistic factors. Mimouni et al. (1998) foundthat broken plurals were produced faster thansound plurals by healthy speakers and by partici-pants with aphasia. This was the case for singularforms of different types of plurals—that is, singu-lars of broken plurals were responded to fasterthan singulars of sound plurals. They claimed thatArabic plural forms have differential processingroutes: broken plurals are accessed as whole wordforms, but sound plurals are decomposed intoword form + suffix.

2. The current study

The failure for any model to account for all the data,and the fact that most of investigations have beendone on English and other Indo-Europeanlanguages, warrants further investigation inArabic, which uses a different morpho-syntax.

Cognitive Neuropsychology 343

Arabic offers the opportunity to study the phenom-enon in question because of its morpho-syntacticfeatures. According to McCarthy (1975, 1981)and Prunet, Beland, and Idrissi (2000), Arabicword forms are composed of two morphemes: aconsonantal root (delivering lexical meaning) anda vocalic pattern (delivering syntactic information—i.e., gender and number). This feature couldenable morphological detachment of the root andpattern, hence separating lexical meaning from syn-tactic information at word form level.

This study investigates morpho-syntactic pro-cessing of regular and irregular inflected formationsof Arabic plurals through qualitative and quantitat-ive analysis, in which vocalic patterns are isolatedfrom consonantal roots. To the best of the research-ers’ knowledge, it is the first investigation ofregular and irregular processing that attempts todetach lexical meaning from morpho-syntacticinformation at a surface level, using a data analysismethod developed for non-concatenativemorphology.

The primary aim of the study is to test whetherArabic sound and broken plurals are processed dif-ferentially or equally. Furthermore, the study hastwo secondary aims; whether Arabic data fulfilsthe assumption of linking agrammatism to regularinflection impairment, and whether the proposedmodels (Joanisse & Seidenberg, 1999; Ullmanet al., 1997) withstand language-specific featuresgenerated by Arabic data.

The dialect in question in the current study isurban Jordanian Arabic (hereafter, Arabic), whichis a sub variety of Levantine Arabic. Arabic hastwo plural types. Duals refer to two items only.Plurals that refer to three or more items aredivided into sound and broken plurals. Duals andsound plurals are formed through inflected suffixa-tion of the singular form of a given noun and aredeemed the regular form. Broken plurals areformed through changing the vocalic pattern ofthe noun and are deemed the irregular form.Examples (1) and (2) present feminine duals andmasculine duals, respectively.

(1) /wərəg-teɪn/ [FEMININE DUAL NOUN]“two papers”

(2) /gələm-eɪn/ [MASCULINE DUALNOUN] “two pens”

Sound plurals are largely confined to nouns thatindicate profession and habit. Nouns such as“engineer”, “teacher”, and “liar” are pluralized inthis way. Examples of sound plurals are providedin (3) and (4):

(3) /mʊdarɪs/ [MASCULINE SINGULARNOUN] “instructor”; / mʊdarɪs-i:n/ [MAS-CULINE PLURAL NOUN] “instructors”

(4) /mʊhəndɪs/ [MASCULINE SINGULARNOUN] “engineer”; /mʊhəndɪs-ʊn/ [MAS-CULINE PLURAL NOUN] “engineers”

Broken plurals are formed by stem change in thevocalic pattern of the noun. This process is depen-dent on two morphological phenomena: the vocalicpatterns of broken plurals and the consonantalroots. Lexical items of Arabic origin consist of aconsonantal root, which is a string of consonants.Most roots in Arabic are made up of three conso-nants, which are referred to as triliteral roots or con-sonantal roots (Prunet, 2006; Zabbal, 2002). A rootcan be notated as C1 C2 C3. Roots are combinedwith vowels that embed themselves between rootconsonants to form a lexical item. The meaningof a given lexical item can be changed by changingthe vowel pattern within the word form. Forexample, the word form /kətəbə/, meaning “hewrote”, is derived from the consonantal root /ktb/,which is the triliteral root for word forms derivedfrom these three consonants. The same triliteralroot /ktb/ is used with a change of vowels for theword form /kɪtæbə/, meaning “writing”. Previousliterature (Soudi, Cavalli-sforza, & Jamari, 2002;Prunet, 2006; Zabbal, 2002) has stated thatbroken plurals are formed in a three-stageprocess. For the sake of describing this process,the singular noun /rəʤʊl/ “man” will be used asan example through the three stages of broken plur-alization process. Stage 1: the root from the under-lying singular noun is selected. Stage 2: the brokenplural pattern—i.e., vocalic pattern—is selected.Stage 3: the selected root is merged with theselected vocalic pattern to form the broken plural

344 T. Khwaileh et al.

of the singular noun /rəʤʊl/ “man”, to become/rɪʤɑ:l/ “men”.

3. Methods

3.1. Participants

Participants were recruited from two hospitals intwo major cities in Jordan. They were first con-tacted through their speech and language therapists;they then received an information sheet about theproject. Prior to entering the study, they signed aconsent form. A confidentiality form was signedby the researcher. Recruitment and participationprocedures were approved within the researchethics processes of both hospitals and the Univer-sity of Sheffield, UK.

Participants were literate native speakers ofurban Jordanian Arabic, over the age of 18 years,and right-handed; they had had normal develop-ment of speech and language prior to their injury.Their aphasia was a result of a single left cerebro-vascular accident (CVA), at least six months priorto their involvement in this study. They presentedwith anomia, and no or mild dysarthria or apraxiaof speech, and no homonymous hemianopia.They had adequate auditory and visual comprehen-sion to enable them to follow test instructions, andnormal or adjusted-to-normal vision and hearing.They did not show signs of other significant neuro-logical or psychiatric disorders. Since the focus ofour investigation is on morpho-syntax, all partici-pants had varying degrees of agrammatism andword-finding difficulties.

Three participants with aphasia—two females(P1 and P2) and one male (P3)—were recruited tothe study. Background information includinginitial diagnosis and medical history were providedby participants’ speech and language therapists andneurologists. Table 1 summarizes their profiles.

3.2. Materials and experimental procedure

Two sets of materials were used to collect data. Thefirst set consisted of a number of aphasia assess-ment subtests, the aim of which was to establish aprofile of each participant’s aphasia and identifythe functional locus/loci of the anomia. Thesecond was an Arabic naming test aimed at Table1.

Backgroundinform

ationforparticipantswith

aphasia.

Participant

Age

(years)

Sex

Hand

Hem

iplegia

Etio

logy

Tim

epost-onset

(months)

Lesionsite

Educatio

n(years)

P1

22F

Right

Hem

iplegiaright

CVA

12L-anteriorandmiddlecerebralartery

15P2

24F

Right

Hem

iplegiaright

CVA

9L-fronto-parietallesion

16P3

62M

Right

Hem

iplegiaright

CVA

12L-parietalarea

12

F=female,M

=male,L=left,C

VA=cerebrovascularaccident.

Cognitive Neuropsychology 345

assessing the production of regular and irregularplurals in Arabic. The researcher administered allassessments.

Participants were assessed individually inspeech and language therapy clinic rooms. Priorto each subtest, participants were informed aboutthe subtest procedure. All instructions werepresented in spoken and written Arabic. Partici-pants were also presented with practice itemsprior to each subtest in order to ensure that theyunderstood what was required of them. If a partici-pant did not understand what was required, furtherinstructions and practice items were given. Partici-pants were informed that they could ask for a breakor end their participation at any time.

3.2.1. Aphasia assessment subtests andneuropsychological profiles

The selection of these tests was constrained by thematerials available for assessment in JordanianArabic in clinics in Jordan at the time of assess-ment. These subtests were taken from twosources: translated subtests of the ComprehensiveAphasia Test (CAT) (Swinburn, Porter, &Howard, 2004) and subtests that have been devel-oped in speech and language clinics in Jordan.The CAT subtests had been translated by cliniciansand were in use in Jordan. They were modified tosuit local cultural and linguistic criteria. In addition,a connected speech sample was recorded from eachparticipant. The sample served as a measure oflexical retrieval and grammatical construction inconnected speech. Transcription, coding, andanalysis of the output was conducted in line withrecommendations suggested by Herbert, Best,Hickin, Howard, andOsborne (2008, pp. 200–202).

The assessment results in Tables 2 and 3 showthe neuropsychological profiles for participants. Allthree participants had anomia and agrammatism.

3.2.1.1. P1’s neuropsychological profile. P1 is aright-handed female who is a native speaker of Jor-danian Arabic and literate; she also speaks Englishas a second language. She was 22 years old at thetime of the study. She was educated in Arabic atschool. Prior to her injury, P1 was a final-yearundergraduate student of information technology.

She suffered a single left cerebrovascular accident(CVA) 12 months prior to her participation in thisstudy, as a complication of surgery for resectionof a carotid body tumour. Her most recent compu-terized tomography (CT) scan, conducted fourmonths post-onset, showed a large area of lowattenuation involving left anterior and middle cer-ebral artery territories. In addition, P1’s CVAresulted in encephalomalacia involving the abovearterial distributions. This encephalomalacia wasthe result of the CVA and was not degenerative,as reported by her neurologist. Her speech andlanguage therapist described a right-sided hemiple-gia and Broca’s aphasia.

P1’s picture naming was outside the normalrange. Her semantic errors included one semantic-coordinate error (/sɪkɪnə/ “knife” → /məlʕəqə/“spoon”), two semantic-associate errors (e.g.,/sɪgərə/ “cigarette” → /dʊxɑ:n/ “smoke”), and onesemantic-visual error (/ʤrəbə/ “sock” → /rɪʤl/“foot”). All phonological errors were phonologicalrelated non-word forms (e.g., /ʔnənɑ:s/ “pineap-ple”→/ʔnɑ:s/ “non-word form”). P1’s lexical retrie-val in the middle five minutes of the conversationwas analysed in line with guidelines from Herbertet al. (2008). A sample of this interview is presentedin Extract 1.

Extract 1: A sample of P1’s conversation withthe researcher (TK)

TK: /ɜɪʃ btɜʕməlɪ ləmə btrɔːħi əl beɪt/?“What do you do when get backhome?”

P1: /ɜɾ (13.8) əʃæg (2.5) əʃɾəb (3.0) kɔljə(6.4) ah/[er – phonological paraphasia – simplepresent verb – feminine singular noun– ah]“er (13.8) [phonological paraphasia](2.5) drink (3.0) college (6.4) ah”

TK: /ʃʊ kəmɑ:n/?“what else?”

P1: /əsħɑ:b(4.3) jəʕnɪ/[masculine broken plural noun – firstperson pronoun – simple present verb]“friends (4.3) I mean”

TK: /ɜɪʃ btɜʕməlʊ mɑʕ bəʕð/?“what do you do together”

P1: /məjə (5.2) əʃɾəb (2.8) ənæ məjə (3.7)kɔlʊ əʃɾəb (3.5)/[noun – simple present verb – first

346 T. Khwaileh et al.

person pronoun – noun – expletivespeech filler – simple present verb]“water (5.2) drink (2.8) I water (3.7)everything drink (3.5)”

P1’s expressive language was non-fluent and effort-ful. She could correctly produce single word forms,with pauses in-between. P1 did not produce well-formed syntactic structures. Phrases were not gram-matical, as they lacked agreement. Her language

did not include subject–verb–object sentences andcomplex sentences. She produced a total of 24single items (17 content word forms and 7 parapha-sias). Content word forms included 9 (53%) nouns,4 (23%) verbs, 3 (18%) adjective, and 1 (6%)adverb. Her errors included 6 (86%) phonologicalparaphasias and a circumlocution (14%). Therewas an abundance of nouns with respect to verbs.All the verbs that P1 produced were in the simple

Table 2. Aphasia assessment results.

Category ofsubtest Subtest

P1 P2 P3

NormMean (range)

Rawscore (%)

Rawscore (%)

Rawscore (%)

Lexicalretrieval

Picture naming (n = 24) 6 0.25 15 0.63 14 0.58 23.3 (21–24)

Inputprocessing

Auditory minimal pairsdiscrimination (n = 10)

4 0.40 10 1.00 8 0.80 Not available

Auditory lexical decision(n = 12)

7 0.58 12 1.00 11 0.92 Not available

Visual lexical decision (n = 15) 14 0.93 14 0.93 10 0.67 Not availableSemanticprocessing

Spoken word to picturematching (n = 15)

14 0.93 14 0.93 13 0.87 14.7 (13–15)

Written word to picturematching (n = 15)

12 0.80 14 0.93 14 0.93 14.9 (14–15)

Outputprocessing

Reading aloud of words(n = 24)

4 0.17 20 0.84 19 0.79 23.7 (22–24)

Repetition of words (n = 16) 13 0.81 16 1.00 12 0.75 15.9 (15–16)Reading aloud of non-words(n = 5)

0 0.00 1 0.20 1 0.20 4.7 (3–5)

Repetition of non-words (n = 5) 1 0.20 3 0.60 1 0.20 4.7 (2–5)Syntacticprocessing

Spoken sentence to picturematching (n = 16)

8 0.50 15 0.94 12 0.75 15.3 (14–16)

Written sentence to picturematching (n = 16)

10 0.62 13 0.81 9 0.56 15.2 (12–16)

Table 3. Summary of results from P1, P2 and P3 conversation samples.

Category Subcategory

Number of items

P1 P2 P3

Speech units N/A 45 113 105Turns Total turns 11 15 17

Substantive turns 3 3 7Minimal turns 8 12 10

Content words excluding paraphasias N/A 17 80 76Number of nouns N/A 9 36 25Errors Circumlocutions 1 2 0

Phonological paraphasias 6 6 7

Cognitive Neuropsychology 347

present tense and inflected for the first persononly. Function word forms were limited to thefirst person pronoun /ənæ/ “I” and the quantifier/kɔl/ “every”. These features indicate that P1 hadagrammatic production (Basso, 2003; Caramazza& Berndt, 1985; Saffran, Berndt, & Schwartz,1989).

Numbers of speech units (verbal and non-verbalvocal utterances), turns (including substantive andminimal turns), content word forms (excludingparaphasias), nouns, and errors were measured inthe middle five minutes of the sample. The majorityof her turns (72%) were lexically empty—that is,minimal turns. Only 28% were substantive turns—turns that included content word forms. Longpauses preceded lexical items. The predominanterror type in P1’s conversation was phonological.All her phonological errors consisted of phonologi-cal related non-word form errors (e.g., /tɪlɪfu:n/ “tel-ephone” → /lɪfu:n/ “non-word form”). Her onlysemantic error was a circumlocution. This perform-ance indicated that lexical retrieval difficulties werepresent in P1’s conversation.

Lexical retrieval difficulties were present in P1’spicture naming and conversation. The presence ofsemantic errors, failures to respond in picturenaming, and semantic circumlocution in conversa-tion indicates a possible semantic deficit, ordamage to the route connecting semantics and thephonological output lexicon (Caramazza & Hillis,1990; Whitworth, Webster, & Howard, 2005). P1may have produced other semantic errors in conver-sation that could not have been noticed by theresearcher, as it is difficult to identify target wordforms in a conversation. The presence of phonologi-cal related non-word form errors in picture namingand conversation indicates damage to the phonolo-gical output lexicon or to processes within phonolo-gical assembly (Whitworth et al., 2005).

P1’s scores on the auditory minimal pairs dis-crimination and auditory lexical decision tasks indi-cate impairment to the auditory input system. Herperformance on the visual lexical decision tasksuggests that visual orthographic analysis wasintact.

P1’s score in spoken word to picture matchingwas within the normal range, which shows thatsemantic processing from spoken input was intact

for this set of items. Her good spoken word formcomprehension was accompanied by deficits inthe auditory phonological analysis and the phono-logical input lexicon. This contrast can be inter-preted as a result of top-down processing fromsemantics in tasks involving pictures (Humphreys,Riddoch, & Price, 1997), which supports the weakor distorted auditory message.

P1’s performance in written word to picturematching is just outside the normal range, whichimplies that semantic processing from visual inputis affected. This was accompanied by good visuallexical decision, which indicates that the source ofher errors on this task cannot be attributed todeficient orthographic input lexicon. Her spokenword to picture matching was within the normalrange, which indicates that semantic processing isintact. This contrast suggests that P1’s errors onthe written word form comprehension resultedfrom impaired access to intact semantics fromintact orthographic input lexicon. Her errors inwritten word to picture matching were two selec-tions of a semantic distractor and one selection ofan unrelated distractor. Semantic distractor selec-tion can result from the degree of visual similaritybetween the semantic distractor and the targetitem. For example, she selected a picture of“shoes” instead of the target picture “boots”.Cole-Virtue and Nickels (2004) maintain thatsemantic distractors in spoken/written word topicture matching tasks share visual properties withthe target items, which could lead to confusion.

P1’s semantic processing was intact for this setof concrete items. The difficulties in lexical com-prehension tasks occurred at levels prior to seman-tic system. These difficulties may be due to inputprocessing rather than to a central semanticdeficit. This suggests that her lexical retrieval diffi-culties did not arise at the semantic level. Hersemantic errors in picture naming were a result ofdamage to the route connecting semantics and thephonological output lexicon (Caramazza & Hillis,1990).

In reading word forms aloud, P1’s score wasoutside the normal range: correct = 17%; phonolo-gical errors = 8%; perseveration = 8%; unrelatedresponse = 13%; no response = 54%. Her phonolo-gical errors resulted in non-word forms (e.g.,

348 T. Khwaileh et al.

/mʊstəʃfə/ “hopsital”→ /mʊʃfə/ “non-word form”).P1’s failures to respond in reading aloud can be aresult of damage to the route connecting semanticsand the phonological output lexicon. If readingbypasses semantics, failures to respond can resultfrom difficulties in accessing the phonologicaloutput lexicon from the orthographic inputlexicon. Her perseverative errors may result fromdifficulty in activating the target response andtherefore retrieve a more available item [see Plautand Shallice (1993) for a connectionist view onperseverations].

In repetition of word forms, P1’s score wasoutside the normal range. Her errors consisted ofall phonological related non-word forms (e.g.,/mɪfɪtæħ/ “key”→/ɪtæħ/ “non-word form”). Herperformance in this task may have been affectedby her impaired auditory input processes and pho-nological assembly. This performance is betterthan her performance in picture naming andreading aloud. This contrast suggests that theprocess of repeating word forms was supportedby direct lexical processing from the phonologicalinput lexicon to the phonological output lexicon,bypassing semantic processing, and by sub-lexicalprocesses simultaneously.

Reading non-word forms is completelyimpaired, as P1 failed to read the presented items.This indicates damage to the orthographic-to-pho-nological-conversion sub-lexical route. In rep-etition of non-word forms, her score was outsidethe normal range. The only non-word form thatP1 could repeat was a monosyllable (/θi:b/“CVC”). This indicates damage to the phonologi-cal-input-to-output-conversion sub-lexical route.P1’s digit and phoneme spans were reduced to 2digits and 2 phonemes, which is below the norm(mean = 6.4, range = 5–7). This suggests that anauditory short-term memory difficulty waspresent. Damage to phonological assembly contrib-uted to her inability to read and repeat non-wordforms, as it is reliant on maintaining the set of pho-nemes in the short-term memory without lexicalsupport.

P1’s phonological errors in all spoken outputtasks and conversation indicate damage to pro-cesses within the phonological output lexicon orwithin phonological assembly. However, all

spoken output tasks were affected. All her phonolo-gical errors in picture naming, reading aloud, andrepetition involved target word forms of three ormore syllables, which suggests damage to phonolo-gical assembly (Butterworth, 1992). Furthermore,her non-word form processing was worse than herword form processing, as is evident from herscores on reading and repetition tasks. Her overallpattern of performance suggests that her phonologi-cal errors arise at phonological assembly levelrather than the phonological output lexicon.

P1’s comprehension of syntactic structures inthe conversational speech sample was relativelywell preserved, as she could understand the ques-tions addressed to her. Her scores in spoken andwritten sentence to picture matching, on the otherhand, were poor. Her scores on both spoken andwritten versions of this subtest were outside thenorm. In spoken sentence comprehension, hererrors included 3 subject–verb–object reversiblesentences, 2 sentences with embedded structures,and 3 reversible sentences with locative preposi-tional phrases. In written sentence comprehension,her errors included 3 subject–verb–object reversiblesentences, 1 sentencewith passive construction, and2 reversible sentences with locative prepositionalphrases. P1 presentedwith impaired comprehensionof syntactic structures. Difficulties in processingthese types of syntactic constructions are key fea-tures of agrammatism at comprehension level(Basso, 2003).

3.2.1.2. P2’s neuropsychological profile. P2 is aright-handed female who is literate and a nativespeaker of Jordanian Arabic; she is monolingual.She was 24 years old at the time of the study. Sheholds an undergraduate degree in education andwas educated in Arabic throughout her schooling.Prior to her CVA, P2 was a primary schoolteacher. Nine months prior to LA’s participation inthis study, she suffered a single CVA resultingfrom an arteriovenous malformation (AVM) onthe middle cerebral artery (MCA) causing a leftsylvian haematoma. Computerized tomography(CT) scan conducted two days post-onset showeda fronto-parietal acute haemorrhage in the areamainly supplied by the MCA. Her speech and

Cognitive Neuropsychology 349

language therapist described a right-sided hemipar-esis and transcortical motor aphasia (TMA).

P2’s performance in picture naming was outsidethe normal range. Her semantic errors included onesemantic-coordinate error (/θəʕləb/ “fox”→/kəlb/“dog”) and one semantic-visual error (/mədxəl/“gate”→/bæb/“door”). All phonological errors werephonological related non-word (e.g., /mʊɣələf/“envelope”→/mʊɣələb/ non-word). P2’s lexicalretrieval in conversation was analysed. A sample ofthis interview is presented in Extract 2. Pauselengths are presented in brackets. The participant’sresponses to the questions are presented with agloss. A translation of each turn is provided.

Extract 2: A sample of P2’s conversation withthe researcher (TK)

TK: /mətə təxærʤtɪ/?“When did you graduate?”

P2: [audible breath] sənə ləʔ θni:n (2) jəʕnɪgəbʌl (1) sənteɪn[feminine singular noun – negationparticle – cardinal number – present-tense verb – feminine dual noun]“a year no two (2) I mean (1) twoyears.”

TK: /ʃʊ btɪðəkærɪ ʕən həðək el ju:m/?“What do you remember from thatday?”

P2: /ʌh ħelʊ sæħbætɪ eʤɪw (1) ʃʊ esemhæ?ər (2) ər (1) əri:ʤ wə kəmæn səħər/[speech filler – adjective – feminineplural noun – past-tense verb – inter-rogative particle – feminine singularnoun – word fragment – word frag-ment – proper noun – conjunctive par-ticle – proper noun]“ah nice, my friends come what’s theirnames?” “Ar Ar Areej and Sahar”

P2’s expressive language was non-fluent and effort-ful and contained pauses. Her responses to ques-tions were interrupted with word-findingdifficulties. The majority of her utterances weresimplified grammatical structures. Single wordsand two-word phrases formed the vast majority ofher utterances. The most complex syntactic struc-tures she produced were subject–verb–object sen-tences, which were grammatical. She did notproduce complex structures such as passive con-structions. Nouns were abundant over verbs. Sheproduced 80 content word of which 45% were

nouns, 25% were verbs, and the remainder wereadjectives and adverbs. Simple present-tenseverbs were more common than past-tense verbs,which formed 10% of verbs produced in total. Pro-duction of function word was restricted to the nega-tion particle /læ/ “no”, the interrogative particle /ʃʊ/“what”, and the conjunctive particle /wə/ “and”.These features indicate that P2 had agrammatic pro-duction (Basso, 2003; Caramazza & Berndt, 1985;Saffran et al., 1989).

Numbers of speech units (verbal and non-verbalvocal utterances), turns (including substantive andminimal turns), content words (excluding parapha-sias), nouns, and errors were measured in themiddle five minutes of the sample. The vastmajority of P2’s turns (80%) were lexically empty—that is, minimal turns. Only 20% were substan-tive turns—that is, included content words. Hererrors consisted of semantic circumlocutions (e.g.,/əʤæmɪə/ “the university” → /bnʌdrʊs mʊhæðərə/“study lecture”) and phonological related non-words (e.g., /həflə/ “party” → /həfl/ → non-word).This performance indicates that lexical retrieval dif-ficulties were present in her conversation. It is poss-ible that P2 had greater difficulties in lexicalretrieval than her conversation sample showed, asit is difficult to identify target words unless the par-ticipant provides sufficient information for theresearcher to identify the target word. In confronta-tional naming tasks, the target word is known, andtherefore it is easier to identify errors (Perkins,1995). P2’s performance in picture naming andconversation indicates that she had lexicalretrieval difficulties. The presence of semanticerrors and failures to respond indicates a possiblesemantic deficit, or damage to the route connectingsemantics and the phonological output lexicon(Caramazza & Hillis, 1990; Whitworth et al.,2005). The presence of phonological related non-word errors indicates possible damage to the phono-logical output lexicon or phonological assembly(Whitworth et al., 2005).

P2’s scores on the auditory minimal pairs dis-crimination and auditory lexical decision tasks indi-cate that auditory phonological analysis andphonological input lexicon processes were preserved.Her performance on the visual lexical decision tasksuggests an intact orthographic input lexicon. These

350 T. Khwaileh et al.

results show that auditory and visual inputs to seman-tics were preserved.

P2’s scores in spoken and written word topicture matching are within the normal range. Shemade one error in each test, both of which involvedselection of the semantic distractor. P2’s perform-ance on the input tasks and lexical comprehensiontasks indicates that semantic processing fromspoken and written inputs was preserved, at leastfor this set of items. This suggests that her lexicalretrieval difficulties are unlikely to be due to adeficit in semantic processing. Similar participantspresenting with semantic errors and circumlocu-tions in the absence of a central semantic deficithave been reported in previous studies (e.g., RGBand HW: Caramazza & Hillis, 1990; MH: Herbert& Best, 2010; GM and JS: Lambon Ralph, Sage,& Roberts, 2000).

In reading aloud of words, P2 scored just belowthe normal range. Her inaccurate responses were aphonological error (/mʊstəʃfə/ “hospital” → /məʃfə/non-word), a perseveration, and two failures torespond to the stimulus. Her phonological error indi-cates damage to the phonological output lexicon orprocesseswithin phonological assembly.Her failuresto respond can be a result of damage to the route con-necting semantics and the phonological outputlexicon. However, if reading bypasses semantics,these errors can result from damage to the route con-necting the orthographic input lexicon and the pho-nological output lexicon. Her perseverative errormay have resulted from difficulty in activating thetarget response, causing her to retrieve a previouslymentioned noun, which was more available thanthe target response. A connectionist interpretationof perseverative errors is provided by Plaut and Shal-lice (1993).

In word repetition, P2 repeated all the presentedwords correctly, which contrasts with her perform-ance on the other two spoken word production tasks(picture naming and reading aloud) and conversa-tion. This contrast suggests that the process ofrepeating words is supported by direct lexical pro-cessing from the phonological input lexicon to thephonological output lexicon, bypassing semanticprocessing, and by sub-lexical processes.

P2’s performance in reading aloud and repetitionof words is better than her performance in picture

naming. This suggests impaired access to the phono-logical output lexicon from semantics rather than adeficit within the phonological output lexicon (Whit-worth et al., 2005). The presence of phonologicalrelated non-word errors in naming and readingaloud indicates damage to the phonological assem-bly. The contrasting good performance in word rep-etition suggests that processes within phonologicalassembly may be partially damaged; when theoutput is constrained by the input stimulus in theform of auditory word, processing proceedswithout errors. However, errors occur when there isno constraint (such as lexical retrieval in picturenaming and conversation), or limited constraintrequiring trans-coding from graphemes to phonemes(reading aloud).

P2 scored outside the normal range on readingaloud of non-words. The source of this impairmentcannot be impaired visual orthographic analysis, asis evident from her performance on the visuallexical decision task, but it does indicate damage tothe orthographic-to-phonological-conversion sub-lexical route. Her repetition of non-words wasoutside the normal range. This performance indi-cates damage to the phonological-input-to-output-conversion sub-lexical route. Auditory phonologicalanalysis was good, as is evident from her perform-ance on the auditory minimal pairs discrimination.P2’s digit and phoneme spans were reduced to 3digits and 2 phonemes, which is outside the norm(mean = 6.4, range = 5–7). In addition to the possibledamage to sub-lexical routes, her difficulties in pro-cessing non-words were caused by damage withinphonological assembly, as it is reliant onmaintainingthe set of phonemes in the short-term memorywithout lexical support.

P2’s comprehension of syntactic structuresduring conversation was relatively intact; she didnot have difficulties in understanding questions.Furthermore, her performance was within thenormal range in spoken and written sentence compre-hension. P2 did not have problems comprehendingsentences with reversible structures or passive con-structions. The data indicate that P2’s input proces-sing of syntax was intact, for the current set of items.

3.2.1.3. P3’s neuropsychological profile. P3 is a62-year-old right-handed male who is a native

Cognitive Neuropsychology 351

speaker of Jordanian Arabic and literate; he ismonolingual. Prior to his injury, P3 was a retiredassistant engineer. He had been formally educatedup to secondary school; he had left when he was17 years old. Arabic was the language of educationthrough his schooling. Twelve months before hisinvolvement in this study, P3 had suffered asingle cerebrovascular accident (CVA) in the leftcerebral hemisphere. His CVA resulted in a hypo-dense lesion in the left parietal region representingan ischemic infarct, according to his most recentcomputerized tomography (CT) scan conductedeight months post-onset. Prior to his CVA, P3 hada history of hypertension, diabetes mellitus, andischemic heart disease, which are believed to havepredisposed him to CVA. The speech and languagetherapists reported that P3 had a right-sided hemi-paresis and transcortical motor aphasia.

P3’s performance on the picture-naming taskand conversation sample was examined. Responsecategories included correct responses (58%), pho-nological errors (26%), neologisms (8%), and noresponses (8%). P3’s performance in picturenaming was outside the normal range. Phonologicalerrors were the predominant error type. His phono-logical errors resulted in non-word productions. Forexample, he produced /nəʕləb/ “non-word” insteadof /θəʕləb/ “fox” and /gæzi:r/ “non-word” insteadof /xænzi:r/ “pig”.

A sample of P3’s conversation is presented inExtract 3. Numbers of speech units (verbal andnon-verbal vocal utterances), turns (including sub-stantive and minimal turns), content words (exclud-ing paraphasias), nouns, and errors were measuredin the middle five minutes of the sample.

Extract 3: A sample of P3’s conversation withthe researcher (TK)

TK: /ʃʊ kʊnt teʃtəɣel/?“What did you work?”

P3: /bel ərð wəl məsæħə (1) kʊnt əsæʕed əl(2) ʃʊ ʔɪsɪmhʌm? (1) beʃtəɣlʊ wəbersʌmʊ/[preposition – noun – feminine singu-lar noun – past-tense verb – particle –masculine plural noun – simplepresent verb – simple present verb]“In the landscape department (1)

helped the (2) their name? (1) theywork and draw”

TK [cuing]: /mʊ/P3: /æh mʊhəndɪs (2) ləʔ əkθər (1)

mʊhəndɪsɪn [slurred speech]/[speech filler – masculine singularnoun – negation particle – adverb –masculine plural noun]“ah engineer (2) no more (1) engineers[slurred speech]”

P3’s expressive language was non-fluent and effort-ful. P3 produced single words, two-word phrases,and some grammatically correct sentences. Hismost complex grammatical structures weresubject–verb–object sentences and sentences withprepositional phrases as complements to verbphrases. Word-finding difficulties were presentthrough his conversation. On one occasion P3 gavephonological and syntactic information about aword he could not retrieve—i.e., /mʊhəndɪsi:n/“engineers”. He stated that it referred to more thanone person and that it was a long word. A phonemiccue was successful on this occasion. He produced 76correct content words: 37 (48%) nouns, 32 (42%)verbs, and 7 (10%) adjectives and adverbs. Allverbs were correctly inflected, and tenses were pro-duced correctly. In addition, P3’s expressivelanguage contained function words; he producedprepositions (e.g., /be/ “in”), the conjunctive particle/wə/ “and”, interrogative particles (e.g., /ʃʊ/ “what”),the negation particle /ləʔ/ “no”, and pronounsembedded in verbs such as /əsæʕed/ “I helped”.There was no evidence of agrammatic production.Lexical retrieval difficulties were present in P3’sconversation. The vast majority of his turns (59%)were minimal turns. Only 41% were substantiveturns. His phonological errors resulted in non-words.

P3’s picture naming and conversation indicate thatlexical retrieval difficulties are present. His expressivelanguage did not include semantic errors, whichsuggests that the semantic system is intact. The predo-minance of phonological errors in picture naming andphonological errors in conversation demonstrate thatdamage to the phonological output lexicon or phonolo-gical assembly is present. Failures to respond in picturenaming indicate impaired access to the phonologicaloutput lexicon from semantic system.

352 T. Khwaileh et al.

P3’s scores on the auditory minimal pairs dis-crimination and auditory lexical decision tasksshow that his auditory input was good. Visualinput processes were compromised, as is evidentfrom his score on the visual lexical decision task.

P3’s performance on the spoken and writtenword to picture matching is within the normalrange. This suggests that semantic processingfrom spoken and written inputs is intact. Thisclaim is supported by the absence of semanticerrors in his picture naming. However, P3’s goodwritten word comprehension is accompanied byimpaired orthographic input. This contrast suggeststhat the weak distorted orthographic message issupported by top-down processing from semanticsin tasks involving pictures (Humphreys et al.,1997).

P3 scored outside the normal range in readingaloud and repeating words. He produced 4 phono-logical errors and a perseveration in reading. Hiserrors in repetition were 3 phonological errors anda failure to respond. All his phonological errorsresulted in non-words (e.g., /mʊstəʃfə/ “hospital”→ /mæʃəfə/ “non-word”). P3’s perseverative errorin reading may have resulted from inability to acti-vate the phonological form of the target word;instead, he produced a previous response, whichwas more available. His failure to respond in rep-etition indicates difficulties in access to the phono-logical output lexicon to retrieve the phonologicalform of the target. P3’s score was outside thenormal range in reading and repeating non-words.This may be explained by the damage to theprocess within phonological assembly, or it mayindicate additional damage to the orthographic-to-phonological-conversion and the phonological-input-to-output-conversion sub-lexical routes. Hisdigit and phoneme spans were reduced to 2 digitsand 2 phonemes, which is outside the norm(mean = 6.4, range = 5–7). This may have contribu-ted to his inability to read and repeat non-words, asthese tasks are reliant on maintaining the set of pho-nemes in the short-term memory without lexicalsupport (Whitworth et al., 2005).

P3’s performance on all spoken output taskswas impaired. His predominant error pattern in alloutput tasks and conversation were phonologicalerrors resulting in non-words as a result of omission

and metathesis of phonemes. His performance innon-word reading and repetition was worse thanhis performance in word reading and repetition.This pattern of performance suggests that his pho-nological errors in spoken output tasks resultedfrom damage to phonological assembly ratherthan the phonological output lexicon (Whitworthet al., 2005).

P3’s comprehension of syntactic structures inconversation was relatively well preserved, as heunderstood the questions addressed to him.However, his scores in spoken and written sentencecomprehension tasks were outside the normalrange. In spoken sentence to picture matching, hiserrors included 1 subject–verb–object non-revers-ible sentence, 1 sentence with embedded structure,and 2 reversible sentences with locative preposi-tional phrases. In written sentence to picture match-ing, his errors included 2 subject–verb–objectreversible sentences, 2 sentences with passive con-struction, and 3 sentences with embedded struc-tures. P3 presented with impaired comprehensionof syntactic structures. Difficulties in processingthese types of syntactic constructions are featuresof agrammatism at comprehension level (Basso,2003).

Lexical processing and speech sample analysesrevealed that all three participants had similar typesof anomia, of varying degrees of severity; the sever-est was P1’s, and the mildest was P3’s. P1’s lexicalprocessing analysis disclosed the source of herword-finding difficulties in picture naming and con-versation. Access to the phonological outputlexicon from the semantic system and impairedphonological assembly were the functional loci atwhich her anomia arose. Her anomia arises ataccess to the phonological output lexicon fromthe semantic system and partial damage to the pho-nological assembly. P3’s anomia was a result ofimpaired phonological assembly and partialdamage to accessing the phonological outputlexicon from semantics. Patterns of performanceof all three participants suggest impairment to thelemma level, and hence difficulties in morpho-syn-tactic processing of words.

Analysis of their sentence processing tasks andspeech samples reveales that agrammatism waspresent in all participants with varying degrees.

Cognitive Neuropsychology 353

P1’s expressive language in conversation showsthat she presented with agrammatic production, asshe had difficulty in formulating well-formed gram-matical constructions. P1 presented with agramma-tism in comprehension, as revealed by herperformance in sentence comprehension tasks.Agrammatism was present in P2’s production asis evident from her conversation. Spoken andwritten sentence comprehensions suggest thatagrammatism was not present at comprehensionlevel. Input to semantics from auditory and visualstimuli was preserved, as is evident in her perform-ance on the auditory and visual input tasks.P3’s expressive language was restricted tosubject–verb–object sentences, as is evident fromhis conversation. Sentence comprehension tasksrevealed that agrammatism was present incomprehension.

3.2.2. Arabic plurals naming subtest

For this test 45 words and their pictorial represen-tations were selected from the Levantine Arabicnormative dataset (Khwaileh, Body, & Herbert,2014). These items were selected according totheir plural type (sound vs. broken). The first setincluded 15 items with sound plurals, and thesecond set included 30 items with broken plurals.Their pictorial representations were presented ona laptop screen in a Microsoft Powerpoint file.Each picture was 320 × 200 pixels. The screenbackground was white. Each picture appeared onthe screen in two different conditions. The 45selected items appeared in two conditions. Thefirst condition was the dual condition, where twopictures of the same object were presented on thescreen at the same time. The second conditionwas the three-plus plural condition, where three ormore pictures of the same item were presented onthe screen. In total, each participant was presentedwith 90 pictorial representations: 45 duals (e.g.,picture of two apples) and 45 sound and brokenplurals (e.g., picture of three apples). Slides wererandomized in order to avoid the effect of locationin the set on picture naming.

In order to ensure that all items were matchedfor significant variables that may affect lexicalretrieval, one-way ANOVA tests were carried out,

which showed that there were no significant differ-ences between the sets with regard to normativenaming latency, F(1, 43) = 0.296, p = .604, age ofacquisition, F(1, 43) = 0.157, p = .444, imageabil-ity, F(1, 43) = 0.240, p = .880, visual complexity,F(1, 43) = 0.004, p = .996, and phoneme number,F(1, 43) = 0.120, p = .138. Stimuli were matchedfor normative naming latency in order to controlfor difficulty of naming, in the absence of frequencyvalues for Jordanian spoken Arabic, followingmethods used for languages with no frequencydata (e.g., Croft, Marshall, Pring, & Hardwick,2011). All Arabic norms were taken from theLevantine normative database (Khwaileh et al.,2014).

Under each presented picture there were twolines to prompt participants to produce twowords. The first word would be the numeral, andthe second word would be the plural form of theword. For example, if there were three repeated pic-tures of a “bird” appearing on the screen at once,they were expected to say /θələθ ʕəsæfi:r/[numeral – masculine plural noun] “three birds”.Producing a numeral referring to the number of pic-tures first would prompt the participant to producethe plural form of the noun.

In the dual condition there was only one linedrawn below the presented pictures, becauseduals in Arabic cannot be preceded by numerals.The number is embedded in the dual suffix /-eɪn/.However, in the three-plus plurals condition, anumeral preceding the noun is essential.

Each participant sat at a comfortable distancefrom the laptop screen. Participants were presentedwith one picture at a time and asked to name theimage aloud as quickly as possible. A Sonyheadset microphone adjusted to 5 cm from each par-ticipant’s mouth was used to detect spokenresponses. The researcher controlled the presen-tation of pictures. Once the participant namedthe pictures appearing on the screen, the researcherclicked a button to move to the next picture.Each stimulus remained on the screen for 20 s. If aparticipant did not name the picture within 20 s ofits presentation, the researcher would move onto the following stimulus. Responses were audiorecorded via a digital recorder (Olympus VN-2100PC).

354 T. Khwaileh et al.

3.3. Coding the data

The data were coded twice by the researcher, toassess intra-rater reliability. The first time was atassessment. The transcribed and coded data werethen recoded by the researcher (r = .912). Thetime between the first coding procedure and thesecond was more than 3 months.

Two main coding systems were developed(Khwaileh, in preparation) lexical accuracy andmorpho-syntactic accuracy; each response wascoded for both factors. The former referred to theretrieval of the lexical stem of the item in question.The latter referred to the retrieval of the appropriatemorpho-syntactic form of the target item. Forexample, a response was coded as lexically accu-rate but morpho-syntactically inaccurate if a partici-pant produced /bəgərə/ [feminine singular noun]“cow” instead of the target noun /bəgəræt/ [femi-nine plural noun] “cows”. However, a responsewas coded as lexically inaccurate and morpho-syn-tactically accurate if a participant produced /kəlbæt/[feminine plural noun] “dogs” instead of the targetnoun /bəgəræt/ [feminine plural noun] “cows”.

4. Results

Five levels of accuracy analysis were conducted:overall correctness (the accuracy of the responseoverall), lexical meaning correctness (the accuracyof consonantal roots regardless of morpho-syntax), morpho-syntax correctness (the accuracyof morpho-syntactic information—gender andnumber—conveyed by the vocalic pattern, regard-less of lexical meaning), morpho-syntactic errortype analysis (e.g., error category), and morpho-syntactic error form analysis (omissions, substi-tutions, insertions).

4.1. Accuracy analysis

Accuracy of regular and irregular plural structureswas compared. Table 4 presents the proportion ofresponses that were correct both lexically andmorpho-syntactically, lexically correct responsesregardless of morpho-syntactic accuracy, andmorpho-syntactically correct responses regardlessof the lexical accuracy. Table4.

Proportions

ofparticipants’accuracy.

Participant

Lexicalandmorpho-syntactic

correctness

Lexicalcorrectness

regardless

ofmorpho-syntax

Morpho-syntactic

correctness

regardless

oflexicalmeaning

Regular

(n=60)

Dualandsoundplurals

Irregular(n

=30)

Brokenplurals

Regular

(n=60)

Dualandsoundplurals

Irregular(n

=30)

Brokenplurals

Regular

(n=60)

Dualandsoundplurals

Irregular(n

=30)

Brokenplurals

P1(proportions)

0.10

0.16

0.58

0.58

0.10

0.33**

P2(proportions)

0.68**

0.40

0.83

0.80

0.73**

0.40

P3(proportions)

0.45**

0.26

0.63

0.53

0.63**

0.36

**Differencestatistically

significant

(p<.05).

Cognitive Neuropsychology 355

Production of plural nouns was impaired in allthree participants. The differences in accuracybetween plural subtypes might be a result of aninability to retrieve the lexical stem, or an inabilityto generate the morpho-syntax. These results indi-cate an effect of morpho-syntactic regularity onaccuracy of plural production. While P2 and P3were more accurate in producing plurals withregular morpho-syntax, P1 was more accurate inirregular plurals.

Statistical analysis of the data presented inTable 4 shows a significant difference betweenregular and irregular plural production for P2(Sign1 test; p < .05) and P3 (Sign test; p < .05),but not for P1 (Sign test; p > .05), when responseswere scored for both lexical and morpho-syntacticaccuracy. There were no significant differences inthe retrieval of lexical stems of regular and irregularplural constructions for P1 (Sign test; p > .05), P2(Sign test; p > .05), or P3 (Sign test; p > .05).Regular and irregular plural productions were com-pared in terms of morpho-syntactic accuracy. Thedifference in retrieval of the morpho-syntaxbetween regular and irregular was significant forP1 (Sign test, p < .05), P2 (Sign test, p < .05), andP3 (Sign test, p < .05). The regularity effectshown in the Table 4 was mirrored in the resultsfor morpho-syntactic accuracy, indicating that forall three participants the dissociation betweenregular and irregular structures was due tomorpho-syntax. The source of this effect wasmorpho-syntactic rather than lexical.

Accuracy analysis of the data showed that aneffect of regularity on plural production waspresent in all participants. P1 was better at produ-cing irregular plurals, and P2 and P3 were better atproducing regular plurals. Accuracy analysissuggests that regular and irregular morpho-syn-tactic structures are processed differentially.This effect was further examined via erroranalysis.

4.2. Morpho-syntactic error type analysis

Error analysis of lexical stems showed that therewas no difference in error types of lexical stemsof regular and irregular forms. Lexical error pat-terns across regular and irregular plurals were

similar in all three participants; therefore,regularity effect on lexical stem errors wasnegligible.

Morpho-syntactic error types and forms wereanalysed to examine whether regular and irregularplurals differ in terms of the types of errorsarising in morpho syntax . Proportions of responsetypes are presented in Table 5.

The lexical substitution of morpho-syntax waspresent in regular and irregular subsets for all par-ticipants. Examples (5) and (6) present instancesof this error type in regular and irregular plurals,respectively.

(5) /ʤəməleɪn/ [masculine dual noun] “twocamels” → /ʤəməl-eθneɪn/ [masculinesingular noun-cardinal numeral] “cameltwo”

(6) /txu:t/ [masculine plural noun] “beds” →/təxt-ərbəʕ/ [masculine singular noun-car-dinal number] “bed-four”

This error pattern indicates that participants couldaccess the lexical stem for this set of items butcould not retrieve the target plural morpho-syntax.All the singular forms of the target noun had pre-served gender. In the above error category, onlynumber inflection was erroneous. It was compen-sated for by adding a cardinal number after thesingular form of the noun. An effect of regularityon this error category was evident in responsesfrom P1, P2, and P3. P2 had a higher proportionof this error in the irregular subset. However, P1and P3 showed a reverse pattern.

The gender inflection error was present in theresponses of P1 and P2 to regular subsets, but notin P3’s responses. It is illustrated in Example (7):

(7) /mʊɣənɪjteɪn/ [feminine dual noun] “twofemale singers” →/mʊɣənjeɪn/ [masculinedual noun] “two male singers”

In the case of broken plurals (irregular), this type oferror was not present. In a broken plural the omis-sion of the gender inflection results in the omissionof the vocalic pattern (as gender is built into thevocalic pattern), which eventually leads the pro-duction of a phonologically illegal word that is

356 T. Khwaileh et al.

not pronounceable. Example (8) illustrates howomission of a vocalic pattern could result in anillegal word.

(8) /səwænɪ/ [feminine plural (broken) noun]“trays” → /swn/ (illegal word)

The number inflection error was present in P1’s andP3’s responses, but not in P2’s. No effect ofregularity was evident on this error category.Examples (9) and (10) illustrate number inflectionerrors in regular and irregular constructions,respectively.

(9) /kɪtæbeɪn/ [masculine dual noun] “twobooks” → /kɪtæb/ [masculine singularnoun] “book”

(10) /xəwætɪm/ [masculine plural noun] “rings”→ /xætɪm/ [masculine singular noun]“ring”

The over-regularization of the morpho-syntax ofthe target word was present in P2’s and P3’sresponses, but not in P1’s. In both cases, therewas an effect of regularity. P2 and P3 producedthis category in response to irregular items only.Examples (11) and (12) illustrate this error type.

(11) /tənəni:r/ [feminine plural (broken) noun]“skirts” → /tənʊræt/ [feminine plural(sound) noun] “skirts (wrong pluralinflection)”

(12) /əħzme/ [masculine plural noun] “belts”→ /ħzəmæt/ [masculine illegal-pluralnoun] “belt + suffix”

The analysis of morpho-syntactic error typesrevealed that two types of errors were governedby the regularity of the morpho-syntax. Whilegender inflection errors occurred in responses toregular structures only, over-regularization of themorpho-syntax occurred in responses to irregularstructures only.

4.3. Morpho-syntactic error forms

Morpho-syntactic errors demonstrated in the pre-vious section were a result of omissions orT

able5.

Proportions

ofmorpho-syntactic

response

types.

Responsecategory

P1

P2

P3

Regular

(n=60)

Dualandsound

Irregular

(n=30)

Broken

Regular

(n=60)

Dualandsound

Irregular

(n=30)

Broken

Regular

(n=60)

Dualandsound

Irregular

(n=30)

Broken

Correct

0.11

0.33**

0.73**

0.40

0.66**

0.36

Lexicalsubstitutionof

morpho-syntax

0.53

0.40

0.07

0.30

0.16

0.06

Num

berinflectio

nerror

0.16

0.07

——

0.01

0.03

Genderinflectio

nerror

0.03

—0.05

——

—Over-regularizatio

nof

morpho-syntax

N/A

—N/A

0.16

N/A

0.30

Other

0.11

0.07

0.07

—0.10

0.13

Noresponse

0.06

0.13

0.08

0.14

0.07

0.12

**Differencestatistically

significant

(p<.05).

Cognitive Neuropsychology 357

substitutions of the target inflection. The currentsubsection examines the effect of regularity onmorpho-syntactic error forms. Error forms werecompared across morpho-syntactic errors toregular and irregular plurals. The number ofmorpho-syntactic errors and the proportion ofomission versus substitution errors for each partici-pant were calculated; they are presented in Table 6.Errors that were classified as “other” and “noresponse” were excluded from the error formanalysis.

There was a difference in error form betweenregular and irregular plural forms. Errors ofregular plurals consisted of omission and substi-tution of the dual or sound plural suffix. However,errors of irregular plurals were substitutions of thevocalic pattern only. Example (13) illustrates anomission error in regular constructions. Examples(14) and (15) show substitution errors of regularand irregular constructions, respectively.

(13) /kɪtæbeɪn/ [masculine dual noun] “twobooks” → /kɪtæb/ [masculine pluralnoun] “book”

(14) /taʊlteɪn/ [feminine dual noun] “twotables” → /taʊlæt/ [feminine plural noun]“tables”

(15) /xəwætɪm/ [masculine plural noun]“rings” → /xætɪm/ [masculine singularnoun] “ring”

Regular errors were mainly omissions, while irre-gular errors were substitutions. It is also importantto note that within errors of regular plural for-mations, omission errors outnumbered substitutionerrors by a considerable amount.

5. Discussion

The data obtained in this study were comparedagainst existing accounts of regular and irregularmorpho-syntactic processing. Accuracy, errortype, and error form analyses suggested differentialprocessing between regular and irregular pluralforms in all participants. The discussion +shedslight on showing how the current data support thedual mechanism account for regular and irregularmorpho-syntax. The degree to which the singlemechanism can account for the data in question isalso addressed.

P1 scored higher in processing irregularmorpho-syntax than regular morpho-syntax. BothP2 and P3 presented a reverse pattern, in whichregular morpho-syntax processing was signifi-cantly higher than irregular. This double dis-sociation suggests that processing of regular andirregular morpho-syntax of words may be governedby independent cognitive processes. The doubledissociation found in the current study appears tobe an effect of (ir)regularity of morpho-syntax,not confounded by other variables, for a numberof reasons. First, all participants were better atretrieving lexical stems than morpho-syntacticforms, as seen through higher scores for lexicalaccuracy than for morpho-syntactic accuracy.Second, the stimuli in the Arabic plural namingsubtest were matched for visual complexity, image-ability, age of acquisition, name agreement, norma-tive naming latency, and length (phoneme number).This indicates that the current results may not havebeen biased by any of these variables. Therefore, itis reasonable to attribute their impairment to regu-larity of morpho-syntax of the presented stimuli,which challenges Chialant and Caramazza’s

Table 6. Number of morpho-syntactic errors, and proportions of substitutions and omissions.

Error form

P1(# morpho-syntactic

errors)

P2(# morpho-syntactic

errors)

P3(# morpho-syntactic

errors)

Regular(44)

Irregular(14)

Regular(8)

Irregular(14)

Regular(11)

Irregular(12)

Substitutions 0.11 1.00 0.12 1.00 0.18 1.00Omissions 0.89 — 0.88 — 0.82 —

358 T. Khwaileh et al.

(1995) model, in which they do not differentiatebetween words based on their morphological fea-tures but merely on their psycholinguistic charac-teristics—that is, frequency of occurrence. Thecurrent data supported Pinker and Ullman’s (2002)proposal in which they maintained that it is the mor-phological system that forms the basis on which thedual mechanism operates. They added that effects ofpsycholinguistic variables could be present withineach route. This view maintains that regular inflec-tion is the predictable form that can be applied tonovel words and invokes a combinatory rule(English past-tense regular form rule: stem + -ed; e.g., print-ed). However, irregular inflections arestored forms that may not be predicted by the formof the stem and may not be applied to novel word(e.g., meet → met). The fact that our stimuli werematched for psycholinguistic variables, yet dis-sociation in retrieval of morpho-syntax was evidentin all three participants, suggests that this dissociationcould not be attributed to psycholinguistic character-istics of the words in question, which is in agreementwith Pinker and Ullman (2002) but not with Chialantand Caramazza’s (1995).

Pinker and Ullman (2002) proposed that agram-matic participants have difficulties in managingrule-based processes, whereas non-agrammaticparticipants do not. Since regular forms rely onrule-based processes, it should be the case thatagrammatic participants have difficulties withregular forms rather than with irregular ones, whilenon-agrammatic participants should be better in pro-cessing regular forms. Consistent with this, P3 didnot have agrammatic production, and he was betterwith regular than with irregular forms. In addition,P1 had agrammatism in production, and, in turn,she showed poorer performance on regular forms.Inconsistent with this, P2 had agrammatism in pro-duction, yet she was better at producing regularthan irregular forms.Both P1 andP2 had agrammaticproduction but showed different patterns of perform-ance. This discrepancy supports previous studies,which point to the fact that linking agrammatism tothe effects of regularity in the dual mechanismaccount needs to be revised (De Diego Balagueret al., 2004; Laiacona and Caramazza, 2004;Shapiro and Caramazza, 2003). These studies havereported participants who showed differential

processing between regular and irregular forms butdid not show the expected pattern of agrammatism.Participants JM and MP (De Diego Balaguer et al.,2004), RC (Shapiro and Caramazza, 2003), andMR (Laiacona and Caramazza, 2004) had agram-matic aphasia but showed poorer performance onirregular forms than on regular ones. De Diego Bala-guer et al. (2004) concluded that “amajor componentof agrammatic production is a deficit in morpho-syn-tactic processing, independently of whether this pro-cessing ultimately involves regular or irregularforms” (p. 212).

Error type analysis suggests that two error typesmay have been governed by the regularity of theplural morpho-syntax. Gender inflection errorsoccurred in regular plural constructions only. Incontrast, over-regularization of morpho-syntaxoccurred in irregular plural morpho-syntax con-structions. Participants produced other error types,but these two types are of interest to this discussion,since their occurrence was governed by the regu-larity of morpho-syntax, as explained below.

Gender inflection errors were present only inregular plurals, which indicated that gender inflec-tion in regular plural formations was separablefrom the stem, unlike irregular plurals, wheregender inflection is built within the vocalicpattern. This inseparability was evident from thelack of gender inflection omissions in P1, P2, andP3’s production of irregular plurals. One couldinterpret this as the result of two different processes:one permitting the omission of gender morpho-pho-nology, and the other not authorizing such omis-sion. However, the fact that the omission ofgender information in irregular forms results in anillegal phonological form suggests that the sameeffect can be present in the case of the single mech-anism system, provided that it follows the phonolo-gical rules of Arabic language. Neither models offeran interpretation of language-specific features likethis one, warranting further modifications ofmodels to fit language-specific features.

An alternative language-specific account pro-vides a better understanding of this error type.The fact that there were no gender inflectionerrors in broken irregular production could be dueto the fact that syntactic information, such asgender, is already embedded in the vocalic patterns

Cognitive Neuropsychology 359

for irregular forms—that is, broken plurals(McCarthy, 1975, 1981). This can be understoodwithin the framework of concatenative versusnon-concatenative morphology (McCarthy, 1975,1981; also described in Prunet et al., 2000). Conca-tenative word form formation refers to linear wordform formation in which affixes are added to thestem to change the inflection or derivation of agiven word (e.g., /məlɪk-æt/ “queen-s”). Regularplurals are formed via a concatenative mechanism.Non-concatenative word form formation isachieved via internal change to the vocalic patternof a given word. Irregular plurals are formed in anon-concatenative mechanism. In non-concatena-tive word form formation, the vocalic patternbears syntactic information, such as gender. Thevocalic pattern in non-concatenative structurescannot be omitted, as this would result in a phono-logically illegal utterance. However, in concatena-tive constructions gender is represented via abound morpheme, which is at risk of omission inimpaired language production. The omission ofthis morpheme results in a real word in its singularform.

The presence of this dissociation in genderinflection errors can be considered a feature ofArabic morpho-syntax. Previous studies investi-gating regular and irregular processing have men-tioned that the dual mechanism approach accountsfor data but fails to explain the presence orabsence of errors resulting from language-specificfeatures. For example, Walenski et al. (2009) inter-preted their Italian data within the dual mechanismapproach; however this approach failed to accountfor an aspect of Walenski et al.’s (2009) results:the lack of regularization errors in the participant’sproduction (e.g., swim→ *swim-ed; Pinker andUllman, 2002). This elucidated importantlanguage-specific differences in regular and irregu-lar processing. The authors attributed this tolanguage-specific features of Italian, such as fre-quency of occurrence of irregular and regular forms.

The second error type of interest is over-regu-larization, where participants produced a regularform instead of the irregular target noun. Accordingto the dual mechanism account, over-regularizationerrors are a result of applying regular inflection toirregular forms when memory fails to supply the

irregular stored form for words. This error typereflects an over-reliance on rule-based mechanismsto compensate for the inability to retrieve irregularforms (Pinker, 1999; Pinker and Ullman, 2002).Participants P2 and P3 presented with loss ofaccess to irregular stored forms and producedover-regularization errors. In contrast, P1 presentedwith agrammatic production and a deficit in thegeneration of regular inflections; she did notproduce over-regularization errors, as impairedrule-based mechanisms may have prevented theseerrors. This dissociation suggests that productionof regular and irregular plurals is governed bytwo independent mechanisms. If a single mechan-ism were involved, it would be possible to haveboth over-regularization of irregular targets, andir-regularization of regular targets.

On an error form level, a dissociation in errorforms of regular and irregular plurals was alsopresent in all participants. Omissions formed thevast majority of errors to regular plurals. In con-trast, substitution was the only error mechanismthat occurred in irregular plural errors. Omissionof inflectional affixes indicates that regularwords can be broken into morphemes that canbe omitted. Substitution errors suggest that irregu-lar forms cannot drop affixes but can substitutethem. This substitution suggests that both conso-nant and vocalic patterns involved in producingan irregular form may be stored, and neither con-sonant roots nor vocalic patterns can be furthersubdivided, but they can be substituted. On theother hand, consonant roots and vocalic patternsof regular forms can be subdivided and affixescan be separated. This strengthens the claim fordifferential representation and processing ofregular and irregular forms.

The findings from the current study are consist-ent with studies from English (e.g., Miozzo, 2003;Newman, Ullman, Pancheva, Waligura, & Neville,2007; Tyler, Stamatakis, & Marslen-Wilson, 2005;Ullman et al., 1997), which reported a double dis-sociation in the processing of English regular andirregular past tense in participants with aphasia.These studies interpreted their results within thedual mechanism approach. The current findingsare also consistent with conclusions from studieson Hebrew (Berent, Pinker, & Shimron, 2002),

360 T. Khwaileh et al.

German (Penke et al., 1999), and Italian (Walenskiet al., 2009). For example, Penke et al. (1999)reported two major findings from German speakerswith aphasia. First, irregular constructions wereselectively impaired, whereas regular formationswere retained. Second, a frequency effect wasobserved for irregular formations but not regularones. The authors interpreted this effect as evidenceon the storage nature of irregular forms. Both find-ings were compatible with the main principles ofthe dual mechanism account (Pinker and Ullman,2002). However, Penke et al. (1999) failed toaccount for the fact that irregular formationsbeing affected after agrammatism violates thedual route assumption that agrammatic participantsretain irregular formations (Pinker and Ullman,2002). Penke et al. (1999) did not offer an interpret-ation for this inconsistency. Furthermore, theauthors did not provide information on the neurop-sychological assessment of participants.

The above discussion has focused on showinghow the current data may be accounted for betterby the dual mechanism. The extent to which thesingle mechanism can account for the data is con-sidered below.

In her study of regular and irregular morpho-syntax in English, Faroqi-Shah (2007) concludedthat models that propose differences in processingregular and irregular morpho-syntax fail toaccount for the whole spectrum of regular mor-phology. She argued that models that make abinary distinction between regular and irregularforms based on past-tense morphology fail toaccount for the presence of preserved progressive–ing or plural –s in the absence of regular mor-phology in the past tense (such as participantsreported in Friedman and Grodzinsky, 1997),suggesting that the dissociation is not a result ofdifferential processing of regular and irregularmorphology. Faroqi-Shah (2007) stated that a dis-sociation between regular and irregular morpho-syntax could be a result of phonological similaritiesand differences rather than morphological features.She argued that English regular past-tense verbshave similar phonological endings, whereas irregu-lars do not. Therefore, a phonological effect couldbe present: for example, a length effect couldresult in dropping the –ed in regular past tenses.

In the current dataset, Arabic data gave insightinto this issue, which English data fail to do. Theregular form of Arabic plural nouns has four differ-ent suffixes, depending on gender and number (dualmasculine: /-eɪn/; dual feminine: /-teɪn/; plural mas-culine: /-ʊn/; plural feminine: /-æt/), unlike theEnglish past tense, which has one regular ending(-ed). Unlike English regular suffixes, Arabicregular suffixes are phonologically distinct, yet adissociation between regular and irregular for-mations was found. This finding supports theclaim that differential processing between regularand irregular morphologies was present andcannot be reduced to phonological similarities, asproposed by Faroqi-Shah (2007), due to the phono-logical distinctiveness of suffixes in Arabic.

A further issue relates to the claim that problemswith regular forms are a result of a phonologicaldeficit, but a deficit in the irregular forms reflects asemantic deficit (Bird et al., 2003; Braber et al.,2005; Joanisse and Seidenberg, 1999; Pattersonet al., 2001). The current data challenge this assump-tion. The functional loci for anomia in P1, P2, and P3were phonological. There was no evidence of acentral semantic deficit. According to Joanisse andSeidenberg’s (1999) assumption, all three partici-pants should process irregular formsmore effectivelythan regular ones.However, P1 showed an advantagefor irregular plurals, whereas P2 and P3 showed theopposite, which is incompatible with the singlemechanism assumption. It is important to mention,however, that the assessment of semantic processingin all three participants was limited to tests that wereavailable at the time of assessment, and thereforemore rigorous testing of semantics would beneeded to eliminate this possibility. The currentdata were consistent with findings from Tyler et al.(2005), who conducted a review of functionalimaging studies and behavioural studies investi-gating regular and irregular morpho-syntax inEnglish. Tyler et al. (2005) concluded that neitherwere regular deficits reduced to phonological impair-ment, nor were irregular deficits reduced to semanticimpairment. Furthermore, participant AW (Miozzo,2003) had intact semantic processing with impairedproduction of irregular nouns and verbs. Thispattern was compatible with participants P2 and P3,who did not have central semantic deficits but had

Cognitive Neuropsychology 361

impaired production of irregular forms, which isinconsistent with the single mechanism assumption(Joanisse & Seidenberg, 1999).

6. Conclusion

The current data showed that the dual mechanismaccount extends to the Arabic morphologicalsystem for word forms, which is a system differentfrom English and from other languages reviewedabove. The dual mechanism theory provides abetter interpretation of the current data than doesthe single mechanism approach. It also shows thatcertain aspects of regular and irregular morpho-syn-tactic processing are governed by the same prin-ciples across languages and are consistent withthe dual mechanism assumptions.

However, this dataset provides evidence offurther issues regarding the dual mechanismaccount. It shows that language-specific featuresmay not be accounted for by the dual mechanismapproach. Language-specific parameters need tobe set to account for data from different languages.The dual mechanism approach could not provideinterpretation to language-specific error types.The lack of gender errors in P1’s, P2’s, and P3’sresponses to irregular stimuli can be attributed tothe non-concatenative nature of Arabic mor-phology, in which gender is embedded in thevocalic pattern of a given noun. Furthermore, thisstudy gives an insight into the nature of processingwithin the dual mechanism approach and agreeswith previous critiques that the link between agram-matism and regular inflection needs to be revised inthe dual mechanism approach, to explain the discre-pancies reported, and to account for language-specific features.

Disclosure statement

No potential conflict of interest was reported by theauthors.

Note1. The Sign test used in this study, unlike the Wil-

coxon Signed-Rank test, does not assume sym-metric distribution of the differences around the

median and does not use the magnitude of thedifference. The Wilcoxon Signed-Rank test needssamples to be equal, but the Sign test does notrequire that.

References

Basso, A. (2003). Aphasia and its therapy. Oxford:Oxford University Press.

Berent, I., Pinker, S., & Shimron, J. (1999). Defaultnominal inflection in Hebrew: Evidence formental variables. Cognition, 72, 1–44.

Berent, I., Pinker, S., & Shimron, J. (2002). Thenature of regularity and irregularity: Evidencefrom Hebrew nominal inflection. Journal ofPsycholinguistic Research, 31(5), 459–502.

Bird, H., Lambon Ralph, M. A., Seidenberg, M. S.,McClelland, J. L., & Patterson, K. (2003).Deficits in phonology and past-tense mor-phology: What’s the connection?. Journal ofMemory and Language, 48, 502–526.

Braber, N., Patterson, K., Ellis, K., & LambonRalph, M. A. (2005). The relationshipbetween phonological and morphological defi-cits in Broca’s aphasia: Further evidence fromerrors in verb inflection. Brain and Language,92, 278–287.

Butterworth, B. (1992). Disorders of phonologicalencoding. Cognition, 42, 261–286.

Bybee, J. L. (1995). Regular morphology and thelexicon. Language and Cognitive Processes,10, 425–455.

Bybee, J. L., & Slobin, D. I. (1982). Rules andschemas in the development and use of theEnglish past tense. Language, 58(2), 265–289.

Caramazza, A., & Berndt, R. (1985). A multicom-ponent deficit view of Broca’s aphasia. In M.L. Kean (Ed.), Agrammatism (pp. 27–63).Orlando: Academic Press.

Caramazza, A., & Hillis, A. E. (1990). Where dosemantic errors come from? Cortex, 26, 95–122.

Cholin, J., Rapp, B., & Miozzo, M. (2010). Whendo combinatorial mechanisms apply in the pro-duction of inflected words? CognitiveNeuropsychology, 27(4), 334–359.

Chialant, D., & Caramazza, A. (1995). Where ismorphology and how is it processed? The

362 T. Khwaileh et al.

case of written word form recognition. In L.Feldman (Ed.), Morphological aspects oflanguage processing (pp. 55–76). Hillsdale,NJ: Erlbaum Press.

Clahsen, H. (1999). Lexical entries and rules oflanguage: A multi-disciplinary study ofGerman inflection. Behavioral and BrainSciences, 22, 991–1013.

Clahsen, H. (2006). Dual-mechanism morphology.In K. Brown (Ed.), Encyclopaedia of languageand linguistics (Vol. 4, pp. 1–5). Oxford:Elsevier.

Cole-Virtue, J. C., & Nickels, L. (2004). Whycabbage and not carrot? An investigation offactors affecting performance on spoken wordto picture matching. Aphasiology, 18, 153–179.

Croft, S., Marshall, J., Pring, T., & Hardwick, M.(2011). Therapy for naming difficulties in bilin-gual aphasia: Which language benefits?International Journal of Language andCommunication Disorders, 48 (1), 48–62.

De Diego Balaguer, R., Costa, A., Sebastián-Gallés, N., Juncadella, M., & Caramazza, A.(2004). Regular and irregular morphologyand its relation with agrammatism: Evidencefrom Spanish and Catalan. Cortex, 40, 157–158.

Faroqi-Shah, Y. (2007). Are regular and irregularverbs dissociated in nonfluent aphasia? BrainResearch Bulletin, 74, 1–13.

Friedman, N., & Grodzinsky, Y. (1997). Tense andagreement in agrammatic production: Pruningthe syntactic tree. Brain and Language, 56,397–425.

Herbert, R., & Best, W. (2010). The role of nounsyntax in spoken word production: Evidencefrom aphasia. Cortex, 46(3), 329–342.

Herbert, R., Best, W., Hickin, J., Howard, D., &Osborne, F. (2008). Measuring lexical retrievalin aphasic conversation: Reliability of a quanti-tative approach. Aphasiology, 22, 184–203.

Humphreys, G. W., Riddoch, M. J., & Price, C. J.(1997). Top-down processes in object identifi-cation: Evidence from experimental psychol-ogy, neuropsychology and functionalanatomy. Philosophical Transactions of theRoyal Society, 352, 1275–1282.

Joanisse, M. F., & Seidenberg, M. S. (1999).Impairments in verb morphology after braininjury: A connectionist model. Proceedings ofthe National Academy of Sciences, USA, 96,7592–7597.

Juola, P., & Plunkett, K. (2000). Why double dis-sociations don’t mean much. In M. Cohen, R.Johnson, & K. Plunkett (Eds.), Exploring cog-nition: Damaged brains and neural networks(pp. 319–327). Philadelphia, PA: PsychologyPress/Taylor Francis.

Khwaileh, T. (in preparation). A theory-drivenanalysis tool for non-concatenative morpho-syntactic disorders: The case of Arabic nouns.

Khwaileh, T., Body, R., & Herbert, R. (2014). Anormative database and determinants oflexical retrieval for 186 Arabic nouns: Effectsof psycholinguistic and morpho-syntactic vari-ables on naming latency. Journal ofPsycholinguistic Research, 43(6), 749–769.

Laiacona, M., & Caramazza, A. (2004). The noun/verb dissociation in language production:Varieties of causes. Cognitive Neuropsychology,21, 103–123.

Lambon Ralph, M. A., Braber, N., McClelland, J.L., & Patterson, K. (2005). What underlies theneuropsychological pattern of irregular >regular past-tense verb production? Brain andLanguage, 93, 106–119.

Lambon Ralph, M. A., Sage, K. E., & Roberts, J.(2000). Classical anomia: A neuropsychologi-cal perspective on speech production.Neuropsychologia, 38(2), 186–202.

Marchman, V. A. (1993). Constraints on plasticityin a connectionist model of the English pasttense. Journal of Cognitive Neuroscience, 5,215–234.

McCarthy, J. J. (1975). Formal problems in Semiticphonology and morphology (Unpublished doc-toral dissertation). Cambridge, MA: MIT.

McCarthy, J. J. (1981). A prosodic theory of non-concatenative morphology. Linguistic Inquiry,12, 373–418.

Mimouni, Z., Kehayia, E., & Jarema, G. (1998).The mental representation of singular andplural nouns in Algerian Arabic as revealedthrough auditory priming in agrammatic

Cognitive Neuropsychology 363

aphasic patients. Brain and Language, 61, 63–87.

Miozzo, M. (2003). On the processing of regularand irregular forms of verbs and nouns:Evidence from neuropsychology. Cognition,87, 101–127.

Newman, A. J., Ullman, M. T., Pancheva, R.,Waligura, D. L., & Neville, H. J. (2007). AnERP study of regular and irregular Englishpast tense inflection. Neuroimage, 34, 435–445.

Patterson, K., Lambon Ralph, M. A., Hodges, J. R.,&McClelland, J. L. (2001). Deficits in irregularpast tense verb morphology associated withdegraded semantic knowledge.Neuropsychologia, 39, 709–724.

Penke, M., Janssen, U., & Krause, M. (1999). Therepresentation of inflectional morphology:Evidence from Broca’s aphasia. Brain andLanguage, 68, 225–232.

Perkins, L. (1995). Applying conversation analysisto aphasia: Clinical implications and analyticissues. European Journal of Disorders ofCommunication, 30, 372–383.

Pinker, S. (1999).Word forms and rules. New York,NY: Basic Books.

Pinker, S., & Ullman, M. (2002). The past andfuture of the past tense. Trends in CognitiveScience, 6, 456–463.

Plaut, D. C., & Shallice, T. (1993). Perseverativeand semantic influences on visual objectnaming errors in optic aphasia: A connectionistaccount. Journal of Cognitive Neuroscience, 5,89–117.

Prunet, J. F. (2006). External evidence and thesemitic root. Morphology, 16, 41–67.

Prunet, J.-F., Beland, R. & Idrissi, A. (2000). Themental representation of Semitic words.Linguistic Inquiry, 31(4), 609–648.

Rodríguez-Fornells, A., Münte, T. F., & Clahsen,H. (2002). Morphological priming in Spanishverb forms: An ERP repetition priming study.Journal of Cognitive Neuroscience, 14, 443–454.

Rumelhart, D. E., McClelland, J. L., & the PDPresearch group. (1986). Parallel distributed pro-cessing: Explorations in the microstructure ofcognition (Vol. I). Cambridge, MA: MIT Press.

Saffran, E. M., Berndt, R. S., & Schwartz, M. F.(1989). A quantitative analysis of agrammaticproduction: Procedure and data. Brain andLanguage, 37, 440–479.

Say, T., & Clahsen, H. (2002). Word forms, rulesand stems in the italian mental lexicon. In S.Nooteboom, F. Weerman, & F. Wijnen (Eds.),Storage and computation in the languagefaculty (pp. 93–129). Dordrecht: Kluwer.

Sereno, J. A., & Jongman, A. (1997). Processing ofEnglish inflectional morphology. Memory &Cognition, 25, 425–437.

Shapiro, K., & Caramazza, A. (2003). Grammaticalprocessing of nouns and verbs in left frontalcortex? Neuropsychologia, 41, 1189–1198.

Soudi, A., Cavalli-sforza, V., & Jamari, A. (2002).The Arabic noun system generation. InProceedings of the International Conferenceon Arabic Processing, Manouba University,Tunisia.

Stockall, L., & Marantz, A. (2006). A single route,full decomposition model of morphologicalcomplexity: MEG evidence. The MentalLexicon, 1, 85–123.

Swinburn, K., Porter, G., & Howard, D. (2004).The comprehensive Aphasia test. Hove:Psychology Press.

Tyler, L. K., Stamatakis, E., & Marslen-Wilson, W.(2005). Differentiating the neurallanguage system: Contrasting the regular andirregular past tense. Brain & Language, 95,172–173.

Ullman, M. (2001). A neurocognitive perspectiveon language: The declarative/proceduralmodel. Nature Reviews Neuroscience, 2, 717–726.

Ullman, M. T., Corkin, S., Coppola, M., Hickok,G., Growdon, J. H., Koroshetz, W. J., &Pinker, S. (1997). A neural dissociation withinlanguage: Evidence that the mental dictionaryis part of declarative memory, and that gramma-tical rules are processed by the proceduralsystem. Journal of Cognitive Neuroscience, 9,266–276.

Walenski, M., Sosta, K., Cappa, S., & Ullman, M.(2009). Deficits on irregular verbal morphology

364 T. Khwaileh et al.

in Italian-speaking Alzheimer’s diseasepatients. Neuropsychologia, 47, 1245–1255.

Whitworth, A. B., Webster, J., & Howard, D.(2005). A cognitive neuropsychologicalapproach to assessment and intervention inaphasia: A clinician’s guide. Hove, UK:Psychology Press.

Yang, C. D. (2000). Knowledge and learning innatural language (Unpublished doctoral disser-tation). MIT, Cambridge, MA.

Zabbal, Y. (2002). The semantics of number in theArabic noun phrase (Unpublished MasterThesis). University of Calgary.

Appendix 1: Lexical accuracy coding system(Khwaileh, in preparation)

1. Correct response: this category was scoredwhen participants produced the target response.1.1. Correct response in slurred speech: this

subcategory was scored when the participantproduced the target response in slurredmanner of speech.

2. Visual error: when participants produced aninaccurate response that shared one or morevisual features with the target response. This cat-egory consisted of two subcategories:2.1. Visual error where participants give a

name of a similar object, such as saying/tɪlɪfɪzjʊn/ “television” instead of/kəmbjʊtər/ “computer”

2.2. Visual error due to a visual distractor inthe presented picture, such as saying a/ʤaɪbə/ “pocket” for a picture of “trouserswith pockets”

3. Semantic error: included inaccurate responseswhere the response shares one or more semanticfeature/s with the target picture. This categoryconsisted of seven subcategories:3.1. Semantic super-ordinate error: pro-

duction of a semantically related error thatis super-ordinate to the target response,such as producing /haɪwæn/ “animal”instead of /xərʊf/ “lamb”.

3.2. Semantic coordinate error: when partici-pants produced a semantically coordinateresponse to the target response, such as pro-ducing /tʊfæhə/ “apple” instead of /mɔ:ze/“banana”.

3.3. Semantic subordinate error: when partici-pants produced a name of a subordinateobject to the target one, such as producing

/hɪsæn/ “horse” instead of /haɪwæn/“animal”.

3.4. Semantic associate error: production of aresponse that is associated to the targetresponse, such as producing /dʊxɑ:n/“smoke” instead of /sɪgərə/ “cigarette”.

3.5. Semantic circumlocution error: pro-duction of a description of the target wordform rather than producing the target wordform itself. This included descriptionswith a minimum of one content wordform. For example, a participant wouldproduce an utterance like /btɪtgəʃər/ “youpeel it” instead of saying /bʊrtʊgæl-ə/“orange”.

3.6. Semantic and visual error: when partici-pants produced an inaccurate response thatshares semantic and visual features withthe target word form, such as producing/leɪmʊn/ “lemon” instead of /bʊrtʊgælə/“orange”.

3.7. Semantic and phonological error: whenparticipants produced an inaccurateresponse that shared semantic and phonolo-gical (share 50% or above of the phonemesof the target response) features with thetarget response, such as producing /hɪmær/“donkey” instead of /hɪsæn/ “horse”.

4. Phonological error: this category included erro-neous responses where the target and the erro-neous response share 50% or more phonemes;for example the participants would say cutinstead of cup. This included three errorsubcategories:4.1. Phonological related real word form:

when participants produced a phonologicalerror that is a real word form, such as produ-cing /kətəbə/ “he wrote” instead of /kɪtæb/“book”.

4.2. Phonological related non-word form: pro-duction of a phonological error that resultedin a non-word form, such as producing/gələd/ “non-word form” instead of /gələm/“pen”.

4.3. Partial production of the target wordform: production of one syllable or part ofthe target word form, such as producing/fʊn/ instead of /tɪlɪfʊn/ “telephone”.

5. Other error: This category included responsesthat did not fit within any of the categoriesabove. This included three subcategories:5.1. Unrelated word form: this subcategory

was scored if participants produced a realword form that is visually, semanticallyand phonologically unrelated to the targetresponse, such as producing /mɪsmær/“nail” instead of /wərəg/ “paper”.

Cognitive Neuropsychology 365

5.2. Unrelated non-word form: production of anon-word form that is phonologically unre-lated to the target response, such as produ-cing /kəbɜ:l/ “non-word form” instead of/fær/ “mouse”.

5.3. Unintelligible response: production of anintelligible response instead of the targetresponse.

6. Morpho-syntactic error: production of thetarget consonantal root with a morpho-syntactic error. This included two mainsubcategories:6.1. Inflectional error: This subcategory was

scored if a participant’s inaccurate responsewas presented with an inflectional error.This was scored if the incorrect number orgender inflections were present, such as pro-ducing /kʊtʊb/ [plural noun] “books”instead of /kɪtæb/ [singular noun] “book”or /mʊmərɪð/ [masculine noun] “malenurse” instead of /mʊmərɪðə/ [femininenoun] “female nurse”.

6.2. Derivational error: this subcategory wasscored if the participant’s inaccurateresponse was presented with a derivationalerror, such as producing an adjective or averb derived from the same consonantalroot of the target response. An example ofthis would be producing / bʊrtʊgæl-i/ [adjec-tive] “orange-adjective” instead of/bʊrtʊgæl-ə / “an orange”.

7. No response: this category was scored when par-ticipants took more than 20 s (from the momentthe stimulus was presented) to respond.

Appendix 2: Morpho-syntactic accuracycoding system (Khwaileh, in preparation)

1. Correct: this category was scored if participantsproduced the target response in full.

2. Error response: included responses other thanthe target response. It included twosubcategories:2.1. Morpho-syntactic error: this subcategory

included responses where participants pro-duced the target lexical item with incorrectmorpho-syntax. It incorporated:2.1.1. Lexical substitution of the morpho-

syntax of the target response: Itincluded producing the uninflectedsingular form of the target noun fol-lowed by a lexical cardinal numberinstead of producing a head nouninflected for number as a result ofomission to the target number inflec-tion. Therefore, the number inflectionwas substituted with a lexical item

representing the number. Forexample:

. /hɪmær eθneɪn/

. [masculine singular noun numeral]

. “donkey two”instead of producing

. /hɪmær-eɪn/

. [masculine-dual noun]

. “two donkeys”2.1.2. Gender inflection error: omission

of the feminine gender inflectionfrom the target noun which resultsin a masculine noun such asproducing:

. təbɪbeɪn/

. [masculine dual noun]

. “two male doctors”instead of

. /təbɪbteɪn/

. [feminine dual noun]

. “two female doctors”2.1.3. Number inflection error: responses

with omissions or substitutions ofnumber inflection in a giventarget noun. A singular or dual formis substituted for a target pluralnoun, or a singular or plural is substi-tuted for a target dual noun. Forexample:

. /kɪtæb-eɪn/

. [dual masculine noun]

. “two books”instead of

. /kʊtʊb/

. [plural masculine noun]

. “books”2.1.4. Over-regularization error: included

responses that were produced with aregular inflection instead of the orig-inal irregular one. For example,broken plurals (irregular inflection)that were produced as sound plurals(regular inflection) were included inthis category. For example:

. /tənʊr-æt/

. [feminine plural (sound) noun]

. “skirts (wrong plural inflection)”instead of

. /tənəni:r/

. [feminine plural (broken) noun]

. “skirts”3. Other response: this category included

responses that did not fit with any of the abovecategories. These responses were mixed and het-erogeneous in nature. In addition they were notfrequent in a participant’s performance and didnot form a recurrent pattern.

366 T. Khwaileh et al.

4. No response: this category was scored ifparticipants did not produce any utterance within20 s from the moment the stimulus was presented.

Appendix 3: Guidelines for conversationanalysisConversation data; taken from Herbert et al. (2008, p.202)

1. Collecting the sample: The sample of conversa-tion should be as close to everyday reality aspossible. To this end the person with aphasiashould record a conversation between themselvesand one other person who is well known to themand does not have aphasia. The conversationshould take place somewhere that is familiar tothem, usually their home.

2. Type of sample: No restrictions upon choice oftopic should be made. The pair should beinstructed to talk as they would do normally.Speakers should be instructed to avoid perform-ing for the tape recorder or doing anything theywould not normally do. Explicit instructionsmay be needed to enable both speakers to takepart, and to avoid a monologue being producedby either of the two speakers.

3. Recording the data:The data is tape-recorded andthe tape later analysed and transcribed.

4. Length of sample: The conversation sampleshould be at least 10 minutes long. From thissample the middle 5 minutes are used for theanalysis. This may mean starting or ending theanalysis midway through a turn. In this casethe partial turn is counted as one turn.

5. Data analysis: The conversation should be tran-scribed using basic conversation analysis nota-tion, and the analysis carried out from thewritten transcription. This is a lengthy procedurebut provides the most accurate means of analysis.Alternatively, for ease in the clinical setting, theanalysis can be performed directly from the taperecording.

6. Output data: The data from the analysis providesraw scores only. In order to compare pre and post

therapy scores relevant proportions of key vari-ables should be computed.

(A) Number of speech units: Count all word formsand other tokens (such as eh? oh, mm, etc.).Treat contractions such as can’t as one unit.Include part word form responses and falsestarts/revisions. Do not include sections ofunintelligible speech.

(B) Turn taking: B1 A turn is a contribution to theconversation. The end of one turn comeswhen the other person starts to talk, or alengthy pause occurs in which neither persontalks. B2 A substantive turn is defined as aturn that contains at least one content wordform, even where this is a repetition of a pre-vious token, or a paraphasia whose target isknown.

(C) Word form retrieval and speech errors: C1Content word forms are defined as nouns,proper nouns, verbs, adjectives, adverbs andnumerals. Adverbs are defined as ending in-ly. When common verbs such as get, have,be, do, know are main verbs they areincluded. If an item is repeated count eachoccurrence of the item separately. Exclusions:Generic terms such as stuff, thing, somethingetc. Modal verbs (can, must, should, might)and auxiliaries be, have, do, will, and shall.Paraphasias. C2 Extract number of nounsfrom above. C3 Semantic paraphasias: thepresumed target and the error are contentword forms. Circumlocutions must containat least one content word form. In bothcases the target must be clear to the conversa-tion partner. Phonological paraphasias: result-ing in both word forms and nonword formswhere the target is apparent. The errorshould contain 50% or more of the targetphonemes in the correct order, or the targetmust contain at least 50% of the error pho-nemes in the correct order. Also includepartial attempts and false starts, e.g., par parpartner. Neologisms: non-word forms wherethe target is not apparent.

Cognitive Neuropsychology 367