Neuropsycholo`ia\ Vol[ 25\ No[ 5\ pp[ 388Ð493\ 0887Þ 0887 Elsevier Science Ltd[ All rights reserved\ Pergamon Printed in Great Britain

9917Ð2821:87 ,08[99¦9[99PII] S9917Ð2821"87#99041Ð7

Clustering and switching on verbal ~uency] theeffects of focal frontal! and temporal!lobe lesions

ANGELA K[ TROYER\�' MORRIS MOSCOVITCH\�$ GORDON WINOCUR\�%MICHAEL P[ ALEXANDER�& and DON STUSS�

� Rotman Research Institute of Baycrest Geriatric Centre^ University of Toronto\ Ontario\ Canada^ $ Department of Psychology\Baycrest Geriatric Centre^ Erindale College\ Mississauga\ Ontario\ Canada^ % Trent University\ Peterborough\ Ontario\ Canada^ and& Braintree Rehabilitation Hospital Network\ Braintree\ Boston University School of Medicine\ Memory Disorders Center\ Boston

Veteran|s Administration Medical Center\ Massachusetts\ U[S[A[

"Received 3 September 0886^ accepted 09 October 0886#

Abstract*We examined the hypothesis that\ on verbal ~uency\ clustering "i[e[ generating words within subcategories# is related totemporal!lobe functioning\ whereas switching "i[e[ shifting between subcategories# is related to frontal!lobe functioning[ Tests ofphonemic and semantic ~uency were administered to 42 patients with focal frontal!lobe lesions "FL#\ 12 patients with unilateraltemporal!lobe lesions "TL# and 44 matched controls[ Performance by FL patients was consistent with our hypothesis] in comparisonto controls\ patients with left!dorsolateral or superior!medial frontal lesions switched less frequently and produced normal clustersizes on both phonemic and semantic ~uency[ Performance by TL patients was not consistent across ~uency tasks and providedpartial support for our hypothesis[ On phonemic ~uency\ TL patients were unimpaired on both switching and clustering[ On semantic~uency\ TL patients were impaired on switching in comparison to controls and left TL patients produced smaller clusters than rightTL patients[ The best indices for discriminating the patient groups\ therefore\ were phonemic!~uency switching "impaired only withfrontal lesions# and semantic!~uency clustering "impaired only with temporal!lobe lesions#[ Þ 0887 Elsevier Science Ltd[ All rightsreserved

Key Words] focal brain lesions^ temporal!lobe lesions^ frontal!lobe lesions^ semantic ~uency^ phonemic ~uency[

Verbal ~uency tasks involve generating words accordingto speci_ed rules within 59 s intervals[ On phonemic~uency\ participants are asked to generate words begin!ning with a speci_c letter\ such as f\ a and s ð0\ 1Ł[ Onsemantic ~uency\ participants are asked to generatewords belonging to a semantic category\ such as animalsð19Ł[ As reviewed subsequently\ phonemic ~uency is oftenthought to be a test of frontal!lobe functioning and sem!antic ~uency a test of temporal!lobe functioning[However\ there is emerging evidence that performanceon these ~uency tasks\ as measured by the total numberof words generated\ is not speci_c to lesions in any singlebrain region and additionally\ is sensitive to di}use braindamage[

Frontal!lobe dysfunction is generally thought to beassociated with impaired phonemic ~uency and relativelyintact semantic ~uency[ Among patients with focal fron!

' Address for correspondence] Angela Troyer\ Departmentof Psychology\ Baycrest Geriatric Centre\ 2459 Bathurst Street\Toronto\ Ontario\ Canada M5A 1E0[ E!mail] troyerÝpsych[toronto[edu[

388

tal!lobe lesions "FL#\ in a majority of empirical studies\impaired performance is obtained on phonemic ~uency\both in comparison to control subjects ð0\ 3\ 4\ 06\ 10\ 11\12Ł and in comparison to patients with posterior corticallesions ð2\ 05Ð07\ 11\ 12Ł[ This _nding\ however\ is notwithout contradiction[ In some studies\ phonemic ~uencyperformance by FL patients has been found to be similarto that of patients with posterior cortical lesions ð00\14\ 17Ł or to patients with di}use brain injury ð11Ł[ Thepatterns of performance by FL patients on semantic ~u!ency are also contradictory[ Although FL patients arenot impaired on semantic ~uency in some studies ð2\ 3\00\ 19\ 17Ł\ impaired levels of performance have beenobtained in others ð4\ 10\ 13\ 14Ł[ Thus\ it is clear thatFL patients exhibit impairment on verbal ~uency tasks\although the speci_c patterns of performance vary some!what between studies[

The cause of the ~uency impairment among FL pat!ients appears to be related to poor initiation and:or ~exi!bility of search and retrieval processes[ Consistent withthis idea\ a patient with bilateral frontal!lobe lesions\in comparison to age!matched controls\ generated fewer

A[ K[ Troyer et al[:Verbal ~uency and focal brain lesions499

words on a standard uncued version of semantic ~uency"i[e[ animals# but improved to the level of controls whensemantic subcategory cues were provided "e[g[ farm ani!mals and jun`le animals# ð13Ł[ Additionally\ FL patients\in comparison to patients with posterior lesions\ per!formed more poorly on a ~uency task requiring alter!nation between animals and s words\ despite no groupdi}erences when animals and s words were generatedseparately ð17Ł[ Both of these studies suggest a speci_cdi.culty switching between categories or subcategories[Additionally\ the ~uency impairment among FL patientsdoes not appear to be due to a failure to use a strategy ofgenerating words within subcategories[ On phonemic andsemantic ~uency\ patients with frontal tumors and pat!ients with posterior tumors did not di}er in the pro!portion of words produced within clusters of three ormore words\ in the size of the largest cluster\ or in thenumber of clusters produced ð01\ 02Ł[

Temporal!lobe dysfunction\ in contrast to frontal dys!function\ is generally thought to be associated with imp!aired semantic ~uency and relatively intact phonemic~uency[ Presumably\ this impairment re~ects temporal!lobe related semantic!memory de_cits[ However\ ~uencystudies of patients with focal temporal!lobe lesions "TL#have been somewhat contradictory[ TL patients per!formed better than FL patients on phonemic ~uency ð2\08\ 12Ł\ but were no more impaired than FL patients onsemantic ~uency ð2\ 19Ł[ Patients with focal temporal!lobe atrophy\ or {{semantic dementia||\ were mildlyimpaired on phonemic ~uency and severely impaired onsemantic ~uency ð8Ł[ Additionally\ patients who hadundergone temporal lobectomy for the control of epilepsywere impaired on both semantic and phonemic ~uencyð03\ 04Ł[ Therefore\ the speci_c relationship betweenlocation of focal brain lesion and performance on pho!nemic and semantic ~uency is unclear[

The e}ect of lesion laterality\ regardless of locationwithin the hemisphere\ on ~uency performance has beenexamined in a number of studies[ Generally\ patients withunilateral left!hemisphere lesions performed more poorlythan patients with unilateral right hemisphere lesions onboth phonemic and semantic ~uency ð09\ 03Ð06\ 08\ 19\11\ 12\ 17Ł[ Patients with bilateral lesions performed theworst ð0Ł[ In some studies\ patients with right!hemispherelesions performed as well as non!lesion controls ð08\ 19Ł\whereas in others\ patients with right hemisphere lesionswere impaired ð03\ 04\ 12Ł[ Overall\ therefore\ the patientgroups ranging from most impaired to least impaired\respectively\ are those with bilateral lesions\ unilateralleft!hemisphere lesions\ and unilateral right!hemispherelesions[

As reviewed previously\ it is evident that the number ofwords generated on verbal ~uency tests is not necessarilysensitive to lesions in any one particular brain region[ Wehave argued ð15Ł that two important cognitive com!ponents of ~uency performance are clustering and swit!ching[ That is\ optimal ~uency performance involvesproducing clusters of semantically or phonemically

related words and once a subcategory is exhausted\ swit!ching to another[ We have demonstrated that switchingwas speci_cally decreased by divided attention ð15Ł andwas consistently impaired in Parkinson|s disease ð16Ł\providing preliminary evidence that switching is relatedto frontal!lobe functioning[ The purpose of the presentstudy was to examine clustering and switching on ~uencytasks among patients with focal brain lesions[ We hypo!thesized that\ in comparison to controls\ switching wouldbe impaired in patients with frontal!lobe lesions "regard!less of lesion laterality# and clustering would be impairedin patients with temporal!lobe lesions "especially left!temporal lesions#[ Our expectations regarding lateralitywere based on the nature of the ~uency components]switching is a cognitive!shifting task without obviousimplications for lateralization\ whereas phonemic clus!tering is a language!related task and semantic clusteringis a verbal semantic!memory task\ both of which may bemore dependent on the left hemisphere[

Methods

Participants

Participants were patients with focal frontal!lobe or focaltemporal!lobe lesions[ Because of group di}erences in demo!graphic characteristics "i[e[ age#\ each was matched with its owncontrol group[ Demographic characteristics are presented inTable 0[

FL patients were grouped according to lesion site and ~uencyperformance via classi_cation and regression tree analyses\ asreported elsewhere ð14Ł[ All lesions were documented on CT orMRI scans[ Fourteen patients had lesions of the left dorsolateraland:or lenticular striate frontal regions "LDLF group#[ Elevenpatients had lesions of the right dorsolateral and:or lenticularstriate frontal regions "RDLF group#[ Seventeen patients hadsuperior medial frontal lesions from either left\ right\ or bifron!tal damage "SMF group#^ right and left lesions were consideredtogether because of previous evidence of lack of lateralizedfunction in the superior medial frontal regions ð14Ł[ Elevenpatients had inferior medial frontal lesions from either left\right\ or bifrontal damage "IMF group#[ A small subgroup ofpatients "00)# with lesions in more than one frontal regionwere classi_ed according to the maximum lesion area[ Theetiology of frontal pathology included cerebral vascular acci!dent "n � 26#\ tumor "n � 7#\ traumatic brain injury "n � 7#\and lobectomy "n � 0#[ The mean time since lesion onset wasat least 2 months for all patients[ Thirty!seven healthy controlsubjects were matched with the FL patients as closely as possiblefor demographic characteristics[ There were no group di}er!ences in age\ F"3\74# ³ 0\ or sex\ x1 "3\ n � 89# � 5[22\ P × 9[09[However\ there were group di}erences in education\F"3\74# � 3[17\ P � 9[992[

Among the 12 TL patients\ 8 had unilateral left!temporal!lobe lesions "LTL# and 03 had unilateral right!temporal!lobelesions "RTL#[ The etiologies of the temporal!lobe lesions werelobectomy for intractable epileptic seizures "n � 19# and cer!ebral vascular accident "n � 2#[ Lesions were documented onCT or MRI scans or by surgical report[ Results of sodiumamytal testing were available for 08 of the 19 patients withtemporal lobectomy^ all were left!hemisphere language domi!nant[ Eighteen healthy control subjects were matched for demo!graphic characteristics with the TL patients[ There were no

A[ K[ Troyer et al[:Verbal ~uency and focal brain lesions 490

Table 0[ Demographic characteristics of patient groups and controls

Frontal groups Temporal groups

LDLF RDLF SMF IMF Ctl LTL RTL Ctl

n 03 00 06 00 26 8 03 07Age 44[5 44[3 43[2 41[4 43[3 26[3 22[0 25[1

"8[6# "01[5# "00[7# "01[0# "03[3# "7[3# "01[8# "03[9#Education\ years 02[6 00[5 00[1 00[6 02[8 01[5 02[1 02[3

"2[1# "0[8# "2[6# "1[0# "1[2# "0[3# "1[5# "0[4#Percentage female 25) 53) 18) 16) 43) 45) 25) 45)

Note] LDLF � left dorsolateral frontal\ RDLF � right dorsolateral frontal\ SMF � superior medial frontal\ IMF � inferior medialfrontal\ Ctl � controls\ LTL � left temporal lobe\ RTL � right temporal lobe[ Standard deviations are shown in parentheses[

signi_cant di}erences between the LTL\ RTL and controlgroups in age\ F"1\27# ³ 0\ sex\ x1 "1\ n � 30# � 0[30\ P � 9[49\or education\ F"1\27# ³ 0[

Procedures and scoring

Tests of phonemic ~uency "i[e[ FAS test# ð0\ 1Ł and semantic~uency "i[e[ animals# were administered on an individual basis[Sixty seconds were allotted for each of the three phonemic trialsand one semantic trial[ Three scores were obtained on each~uency test] "a# total number of words generated\ excludingerrors and repetitions^ "b# mean cluster size^ and "c# number ofswitches[ Detailed scoring rules for cluster size and switcheshave been provided elsewhere and produce scores with highinterrater reliabilities "i[e[ r × 9[84# ð15Ł[ Brie~y\ on phonemic~uency\ clusters consisted of successively generated words thatbegan with the same _rst two letters "e[g[ arm and art#\ di}eredonly by a vowel sound "e[g[ foot and fat#\ rhymed "e[g[ ~ightand fright#\ or were homonyms "e[g[ sail and sale#[ On semantic~uency\ clusters consisted of successively generated wordsbelonging to the same semantic subcategory\ such as farm ani!mals\ pets\ aquatic animals\ or insects[ Cluster size was countedbeginning with the second word in each cluster and mean clustersize was calculated for the phonemic test and for the semantictest[ Switches were calculated as the number of transitionsbetween clusters\ including single words\ for the phonemic andsemantic tests[ The raw number of switches was utilized ratherthan correcting for the number of words generated "i[e[ a pro!portion score#\ because this index best represented the behav!iour of interest[ That is\ optimal ~uency performance\ asmeasured by the raw number of correct words generated\requires the use of a large number of di}erent subcategories\which is achieved by frequent switching[ Optimal ~uency per!formance would not be achieved with a high proportion swit!ching score but a low raw switching score "e[g[ 3 switches outof 4 words#[

Analyses

To correct for group di}erences in education among the FLpatient and control groups\ ~uency performance was analysedwith analysis of covariance "ANCOVA#\ with education as thecovariate[ Group di}erences between the TL patient and con!trol groups were conducted with ANOVA[ Post!hoc pairwisecomparisons "Bonferroni|s test# were used to test whether eachpatient group di}ered from its respective control group[Additionally\ we performed one!tailed t!tests on cluster sizebetween the LTL and RTL groups\ based on our directional

hypotheses[ For the overall ANCOVAs\ ANOVAs\ and t tests\an a level of 9[94 was used^ for the multiple post!hoc pairwisecomparisons\ Bonferroni!corrected a!levels were used[

Results

Fluency data for the patient and control groups\ onboth phonemic and semantic ~uency\ are presented inTable 1[

Frontal groups

Switching performance by the FL patient and controlgroups was consistent with our hypotheses[ That is\ pho!nemic ~uency switching showed signi_cant group di}er!ences\ F"3\73#�05[90\ P³ 9[990\ with impaired perform!ance by the LDLF group\ F"0\73#�49[82\ P³ 9[990 andthe SMF group\ F"0\73#�10[58\ P³ 9[990\ in com!parison to controls[ Similarly\ semantic ~uency switchingshowed signi_cant group di}erences\ F"3\73#�3[61\P�9[990\ with impaired performance by the LDLFgroup\ F"0\73#�02[48\ P³ 9[990 and the SMF group\F"0\73#�09[21\ P�9[991\ in comparison to controls[

Clustering performance by the FL groups was alsoconsistent with our hypotheses[ There were no groupdi}erences in clustering on either phonemic ~uency\F"3\73#�0[09\ P�9[253\ or semantic ~uency\F"3\73#³ 0[

The number of words generated on phonemic ~uencyshowed overall group di}erences\ F"3\73#�08[79\P³ 9[990\ with impaired performance by the LDLFgroup\ F"0\73#�56[07\ P³ 9[990 and the SMF group\F"0\73#�16[57\ P³ 9[990[ The number of words gen!erated on semantic ~uency also showed overall groupdi}erences\ F"3\73#�01[97\ P³ 9[990\ with impairedperformance by the LDLF group\ F"0\73#�26[96\P³ 9[990\ the RDLF group\ F"0\73#�01[52\ P�9[990\the SMF group\ F"0\73#�02[72\ P³ 9[990 and the IMFgroup\ F"0\73#�7[43\ P�9[993[

A[ K[ Troyer et al[:Verbal ~uency and focal brain lesions491

Table 1[ Fluency data for patient groups and controls

Frontal groups Temporal groups

LDLF RDLF SMF IMF Ctl LTL RTL Ctl

Phonemic ~uencySwitches 01[4 14[2 05[3 14[6 18[5��� 13[9 15[2 17[3

"5[8# "6[8# "6[9# "8[6# "7[9# "09[0# "8[9# "7[9#Cluster size 9[22 9[17 9[27 9[29 9[30 9[38 9[36 9[28

"9[11# "9[07# "9[12# "9[11# "9[07# "9[39# "9[05# "9[18#Words generated 06[1 21[8 11[8 23[1 31[6��� 22[5 27[8 30[0

"8[9# "09[8# "7[9# "00[4# "00[3# "09[3# "09[7# "00[5#

Semantic ~uencySwitches 4[7 5[8 4[8 6[3 8[3��� 6[8 6[2 09[7�

"2[1# "1[2# "1[1# "1[4# "2[5# "0[7# "1[5# "2[8#Cluster size 0[10 0[91 0[23 9[86 0[11 9[80 0[24 0[05$

"9[67# "9[49# "9[66# "9[21# "9[38# "9[43# "9[47# "9[44#Words generated 00[8 03[0 03[3 04[0 19[1��� 04[1 07[7 11[4��

"3[9# "1[6# "4[0# "1[4# "3[7# "2[2# "4[4# "3[6#

Note] LDLF � left dorsolateral frontal\ RDLF � right dorsolateral frontal\ SMF � superior medial frontal\ IMF � inferior medialfrontal\ Ctl � controls\ LTL � left temporal lobe\ RTL � right temporal lobe[ Standard deviations are shown in parentheses[� P ³ 9[94\ �� P ³ 9[90\ ��� P ³ 9[990 for overall analyses^ $ P ³ 9[94 for left vs right temporal groups[

Temporal groups

Findings with the TL patients and controls were lessclear\ but provide some support for our hypotheses[ Onphonemic ~uency switching\ there were no overall groupdi}erences\ F"1\27#³ 0[ On semantic ~uency switching\there were overall group di}erences\ F"1\25#�4[91\P�9[901\ with impaired performance by the RTL groupin comparison to controls\ F"0\25#�7[37\ P�9[995[

On phonemic ~uency cluster size\ there were no overallgroup di}erences\ F"1\27#³ 0\ and there was no di}er!ence between LTL and RTL patients\ t"10#�9[19\P�9[735[ On semantic ~uency cluster size\ there wereno overall group di}erences\ F"1\25#�0[55\ P�9[193^however\ consistent with our a priori hypothesis\ LTLpatients produced smaller clusters than RTL patients\t"08#�0[70\ P�9[932[

For the number of words generated\ there were nogroup di}erences on phonemic ~uency\ F"1\27#�0[27\P�9[152[ There were group di}erences on semantic~uency\ F"1\13#�4[66\ P�9[998\ with impaired per!formance by the LTL group in comparison to controls\F"0\25#�03[48\ P�9[990[

Discussion

Our _ndings provide general support for the idea thatswitching on verbal ~uency is related to frontal!lobe func!tioning\ whereas clustering is related to temporal!lobefunctioning[ The pattern of performance by the FL pat!ients was completely consistent with our hypothesis[ Thatis\ in comparison to controls\ FL patients as a group

switched less frequently and produced normal clustersizes on both phonemic and semantic ~uency[ Per!formance by TL patients was not consistent across~uency tasks and provided partial support for ourhypothesis[ On phonemic ~uency\ TL patients were un!impaired on both switching and clustering[ On semantic~uency\ TL patients were impaired on switching in com!parison to controls and LTL patients produced smallerclusters than RTL patients[ The best indices for dis!criminating the patient groups\ therefore\ were pho!nemic!~uency switching "impaired only with frontallesions# and semantic!~uency clustering "impaired onlywith temporal!lobe lesions#[ This is consistent with pre!vious research ð16Ł in which these same indices dis!criminated between dementia groups[ Phonemic!~uencyswitching was impaired only in Parkinson|s dementia "inwhich frontal and subcortical regions are preferentiallyinvolved# and semantic!~uency clustering was impairedonly in Alzheimer|s dementia "in which temporal andparietal regions are preferentially involved#[ We haveobtained converging evidence\ therefore\ that switchingis related to frontal!lobe functioning whereas clusteringis related to temporal!lobe functioning[

The patients with frontal lesions were not homo!geneous in their ~uency performance\ but\ rather\ showedsubgroup performance di}erences[ Speci_cally\ switchingwas impaired in the LDLF and SMF groups and unim!paired in the RDLF and IMF groups[ This is consistentwith research indicating impaired initiation of behaviourand perseveration in patients with dorsolateral frontallesions ð07Ł and superior!medial frontal lesions "i[e[ sup!plementary motor area and anterior cingulate gyrus# ð5\6Ł[ Presumably\ switching on verbal ~uency tasks wouldrequire initiation of numerous searches through semantic

A[ K[ Troyer et al[:Verbal ~uency and focal brain lesions 492

and lexical memory\ and cognitive ~exibility to rapidlyshift between subcategories[

Although switching was impaired on both ~uency tasksin our FL patients\ switching on semantic ~uency wasalso impaired in our TL patients[ It is possible that thisunexpected switching de_cit resulted from damage topathways connecting the anterior frontal and temporalregions subsequent to anterior temporal lobectomy[Thus\ the switching impairment in our TL patients maybe a result of additional frontal dysfunction[ On the otherhand\ it is also possible that switching on semantic ~uencyis mediated both by frontal and anteriorÐtemporalregions[ It is apparent that the frontal lobes are directlyinvolved in switching\ either for initiating switches or forlocating a new subcategory or both[ The frontal lobesmay also require information indicating that the pointof diminishing returns has been reached in searching aparticular semantic cluster[ Perhaps it is this informationthat is not e.ciently transmitted in the semantic ~uencytask in TL patients[ In contrast\ because phonemic pro!cessing is not as dependent on the anterior temporallobe as semantic processing\ switching was normal in TLpatients on this task[ If the frontal lobes are the repositoryof the switching mechanism itself\ frontal damage shouldlead to de_cits in switching on both tasks\ as seen in ourFL patients[

Although semantic!~uency clustering was impaired inour LTL patients\ somewhat unexpectedly\ phonemic!~uency clustering was not impaired in this group[ This issimilar to previous _ndings that impaired clustering is notalways consistent across ~uency tasks ð15\ 16Ł[ Rather\ itappears that di}erent cognitive processes are involved\ atleast to some degree[ Semantic!~uency clustering maydepend on access to and integrity of semantic stores\whereas phonemic!~uency clustering may depend onfund of word knowledge and the ability to analyse thephonemic characteristics of words[ Thus\ despite the factthat clustering on phonemic and semantic ~uency bothinvolve the ability to classify and group words with simi!lar characteristics\ additional cognitive processes varybetween these tasks\ resulting in the possibility of incon!sistent performance across ~uency tasks[

Our frontal!lesion _ndings share some similarities withthose of Laine ð01\ 02Ł\ who found that patients withfrontal tumors produced normal cluster sizes on verbal~uency in comparison to patients with posterior tumors[They also found no group di}erences in the number ofclusters of three or more words\ which was their measureof {{shifting||\ whereas our FL patients were impaired onswitching when clusters of all sizes\ including singlewords\ were considered[ This suggests that switchingalone\ regardless of the size of clusters between which thepatient is switching\ may be the crucial ability that issensitive to lesions of the frontal lobe[

The lateralizing e}ects we obtained for clustering andswitching were generally consistent with our expec!tations[ We had no a priori reason to expect switching\ acognitive!shifting task\ to be di}erentially a}ected by

right! versus left!hemisphere lesions[ Indeed\ we foundthat switching was signi_cantly impaired in patients withleft!frontal\ superior!medial!frontal\ and right!temporallesions[ On the other hand\ we expected that clusteringwords in phonemic or semantic subcategories\ which is alanguage!related task\ would be more greatly a}ectedby left!temporal lesions[ This is indeed the pattern weobtained for semantic!~uency clustering\ although pho!nemic!~uency clustering was unimpaired in all patientgroups[

The pattern of _ndings in Table 1 may appear to sug!gest that switching and number of words generated arehighly dependent variables\ because group di}erences inthe number of words generated were always associatedwith group di}erences in switching[ However\ additionalresearch has indicated that a decrease in the numberof words generated can be obtained without decreasedswitching but\ rather\ with decreased clustering "i[e[among patients with Alzheimer|s disease# ð16Ł[ Impairedswitching\ therefore\ is not necessary for obtaining adecreased number of words generated[

Consistent with the idea that phonemic ~uency is atest of frontal!lobe functioning\ the number of wordsgenerated on phonemic ~uency was impaired in the FLgroup but not the TL group in comparison to controls[This is not only a test of frontal functioning\ however\ asimpaired performance has also been obtained by patientswith parietal!lobe lesions ð14Ł[ Number of words gen!erated on semantic ~uency\ on the other hand\ was imp!aired in both of our patient groups\ suggesting thatperformance on this test is not speci_c to temporal!lobefunctioning[ This is consistent with a number of previousstudies ð4\ 10\ 13Ł[ Speci_c e}ects\ therefore\ wereobtained for the clustering and switching components\but not for the total number of words generated[

The conclusions drawn in this study are limited byseveral factors\ including small TL patient sample sizes\heterogeneity of lesion size and etiology\ and possibledi}use brain dysfunction in addition to focal lesionsamong the patients with tumors or traumatic braininjury[ Additionally\ because of the early age at seizureonset in the temporal lobectomy groups\ some functionalreorganization may have occurred[ Despite these samplecharacteristics\ which likely reduced the probability of_nding group di}erences\ we obtained signi_cant di}er!ences for most of our critical comparisons[

Acknowled`ement*The present research was supported by agrant from the Medical Research Council of Canada to M[Moscovitch and G[ Winocur and grants from the OntarioMental Health Foundation and Medical Research Council ofCanada to D[ Stuss[ A[ Troyer was supported by the Katz Post!doctoral Research Fellowship[

References

0[ Benton\ A[ L[\ Di}erential behavioral e}ects in fron!tal lobe disease[ Neuropsycholo`ia\ 0857\ 5\ 42Ð59[

A[ K[ Troyer et al[:Verbal ~uency and focal brain lesions493

1[ Borkowski\ J[ G[\ Benton\ A[ L[ and Spreen\ O[\Word ~uency and brain damage[ Neuropsycholo`ia\0856\ 4\ 024Ð039[

2[ Corcoran\ R[ and Upton\ D[\ A role for the hip!pocampus in card sorting< Cortex\ 0882\ 18\ 182Ð293[

3[ Coslett\ H[ B[\ Bowers\ D[\ Verfaellie\ M[ and Heil!man\ K[ M[\ Frontal verbal amnesia] Phonologicalamnesia[ Archives of Neurolo`y\ 0880\ 37\ 838Ð844[

4[ Costello\ A[ de L[ and Warrington\ E[ K[\ Dynamicaphasia] The selective impairment of verbal plan!ning[ Cortex\ 0878\ 14\ 092Ð003[

5[ Degos\ J[!D[\ da Fonseca\ N[\ Gray\ F[ and Cesaro\P[\ Severe frontal syndrome associated with infarctsof the left anterior cingulate gyrus and the head of theright caudate nucleus] A clinico!pathological case[Brain\ 0899\ 005\ 0430Ð0437[

6[ Devinsky\ O[\ Morrell\ M[ J[ and Vogt\ B[ A[\ Con!tributions of anterior cingulate cortex to behaviour[Brain\ 0899\ 007\ 168Ð295[

7[ Grossman\ M[\ A bird is a bird is a bird] Makingreference within and without superordinate cate!gories[ Brain and Lan`ua`e\ 0870\ 01\ 202Ð220[

8[ Hodges\ J[ R[\ Patterson\ K[\ Oxbury\ S[ and Funnell\E[\ Semantic dementia] Progressive ~uent aphasiawith temporal lobe atrophy[ Brain\ 0881\ 004\ 0672Ð0795[

09[ Janowsky\ J[ S[\ Shimamura\ A[ P[\ Kritchevsky\ M[and Squire\ L[ R[\ Cognitive impairment followingfrontal lobe damage and its relevance to humanamnesia[ Behavioral Neuroscience\ 0881\ 092\ 437Ð459[

00[ Joanette\ Y[ and Goulet\ P[\ Criterion!speci_creduction of verbal ~uency in right brain!damagedright!handers[ Neuropsycholo`ia\ 0875\ 13\ 764Ð768[

01[ Laine\ M[\ On the mechanisms of verbal adynamia] Aneuropsycholo`ical study[ Doctoral dissertation\University of Turku\ Turku\ Finland\ 0878[

02[ Laine\ M[ and Niemi\ J[\ Word ~uency productionstrate`ies of neurolo`ical patients] Semantic andphonolo`ical clusterin`[ Paper presented at the 05thAnnual Meeting of the International Neuro!psychological Society\ New Orleans\ LA\ 0877[

03[ Loring\ D[ W[\ Meador\ K[ J[ and Lee\ G[ P[\ E}ectsof temporal lobectomy on generative ~uency andother language functions[ Archives of Clinical Neuro!psycholo`y\ 0883\ 8\ 118Ð127[

04[ Martin\ R[ C[\ Loring\ D[ W[\ Meador\ K[ J[ andLee\ G[ P[\ The e}ects of lateralized temporal lobedysfunction on formal and semantic word ~uency[Neuropsycholo`ia\ 0889\ 17\ 712Ð718[

05[ Miceli\ G[\ Caltagirone\ C[\ Gainotti\ G[\ Masullo\

C[ and Silveri\ M[ C[\ Neuropsychological correlatesof localized cerebral lesions in non!aphasic brain!damaged patients[ Journal of Clinical Neuro!psycholo`y\ 0870\ 2\ 42Ð52[

06[ Miller\ E[\ Verbal ~uency as a function of a measureof verbal intelligence and in relation to di}erent typesof cerebral pathology[ British Journal of Clinical Psy!cholo`y\ 0873\ 12\ 42Ð46[

07[ Milner\ B[\ E}ects of di}erent brain lesions on cardsorting[ Archives of Neurolo`y\ 0852\ 8[

08[ Milner\ B[\ Some e}ects of frontal lobectomy in man[In The frontal `ranular cortex and behavior\ eds J[ M[Warren and K[ Akert[ McGraw!Hill\ New York\0853[

19[ Newcombe\ F[\ Missile wounds of the brain[ OxfordUniversity Press\ London\ 0858[

10[ Owen\ A[ M[\ Downes\ J[ J[\ Sahakian\ B[ J[\ Polkey\C[ E[ and Robbins\ T[ W[\ Planning and spatial work!ing memory following frontal lobe lesions in man[Neuropsycholo`ia\ 0889\ 17\ 0910Ð0923[

11[ Pendleton\ M[ G[\ Heaton\ R[ K[\ Lehman\ R[ A[ W[and Hulihan\ D[\ Diagnostic utility of the Thurstoneword ~uency test in neuropsychological evaluations[Journal of Clinical Neuropsycholo`y\ 0871\ 3\ 296Ð206[

12[ Perret\ E[\ The left frontal lobe of man and the sup!pression of habitual responses in verbal categoricalbehaviour[ Neuropsycholo`ia\ 0863\ 01\ 212Ð229[

13[ Randolph\ C[\ Braun\ A[ R[\ Goldberg\ T[ E[ andChase\ T[ N[\ Semantic ~uency in Alzheimer|s\ Par!kinson|s\ and Huntington|s disease] Dissociation ofstorage and retrieval failures[ Neuropsycholo`y\0882\ 6\ 71Ð77[

14[ Stuss\ D[ T[\ Alexander\ M[ P[\ Hamer\ L[\ Palumbo\C[\ Dempster\ R[\ Binns\ M[\ Levine\ B[ andIzukawa\ D[\ Brain mechanisms of verbal ~uency]The e}ects of focal anterior and posterior brainlesions on initiation\ production\ and strategic per!formance[ Journal of the International Neuro!psycholo`ical Society\ in press[

15[ Troyer\ A[ K[\ Moscovitch\ M[ and Winocur\ G[\Clustering and switching as two components of ver!bal ~uency] Evidence from younger and older healthyadults[ Neuropsycholo`y\ 0886\ 00\ 027Ð035[

16[ Troyer\ A[ K[\ Moscovitch\ M[\ Winocur\ G[\ Leach\L[ and Freedman\ M[\ Clustering and switching onverbal ~uency tests in Alzheimer|s and Parkinson|sdisease[ Journal of the International Neuro!psycholo`ical Society\ 0887\ 3\ 026Ð032[

17[ Vilkki\ J[ and Holst\ P[\ Speed and ~exibility on word~uency tasks after focal brain lesions[ Neuro!psycholo`ia\ 0883\ 21\ 0146Ð0151[