Semantic-Phonemic Discrepancy and its changes over time in Alzheimer’s Disease: Evidence from

Clustering and Switching Analyses

aDept. ofCommunicationDisorders,EWHAWomans University,Seoul,SouthKoreabDept.OfCommunicationSciences&Disorders,TempleUniversity,USA

JeeEun Sunga

SangEun Leea

GayleDeDeb

SeJin Oha

Mi KyungShina

SooEun Leea

Semanticvs.Phonemic

• Verbalfluencytasksarefrequentlyusedtoindexcognitive-linguisticdeclineinneurodegenerativeclinicalpopulationssuchasAlzheimer’sDisease(AD).

• Verbalfluencyhasbeenoftenmeasuredintwodomainsusingthesemanticandphonemicfluencytasks.

SemanticFluencygenerateasmanywordsaspossiblethatbelongtoacertainsemantic

category(e.g.,‘animal’)inagiventimeconstraint(e.g.,‘60seconds’)

PhonemicFluencyproducewordsthatstartwithaspecificlettersuchas‘F’,‘S’,or‘A’

underthesametimeframe.

Confrontationnamingvs.Verbalfluency

Theverbalfluencytaskshavebeenregardedasimposinggreatercognitivedemandsthantheconfrontationnamingtasks(e.g.,

Crawford&Henry,2004;Huffetal.,1986 ).

toinhibitwhattheyhavealreadyproducedforsearchingthenewitemsbykeepingtrackoftheir

behaviors

Confrontationnamingtaskvisualstimulià aideasieraccessto

thesemanticmemory

Duetotheadditionaldemandsofself-directedplanningprocedures,theverbalfluencymeasureshavebeenreportedasbeingmoresensitivetodetectthepresenceofneurodegenerativedisease

Verbalfluencytask

Semanticvs.PhonemeDifferencesinprocessingmechanisms

- successfulactivationofsemanticmemory- generalsemanticknowledgefromthelong-termmemory- (Butters ,Granholm,Salmon,&Grant,1987;Hodges,Salmon,&

Butters ,1992)

- accesstotheorthographicandphonemicrelatedness(Brin al.,2010;Weakleyetal.,2014).

- Inhibitionoftheautomaticactivationofwordmeaningsbyprimarilyrelyingonphonemerepresentations.

SemanticFluency

PhonemicFluency

Semanticvs.PhonemeDifferencesinneuralsubstrates

• Temporallobe:relatedtosemanticmemory

• Semantic<Phonemic:ADpatientswithtemporalatrophy

• Frontallobe:inhibitorycontrol,self-directedplanning

• Phonemic<semantic:Participantswithfocalfrontallesions

SemanticDisadvantagePhonemicDisadvantage

Henry&Crawford(2004)AMeta-AnalyticReviewofVerbalFluencyPerformanceFollowingFocalCorticalLesions,Neuropsychology,18(2),284-295

Clusteringvs.Switching

• Toquantifyperformanceontheverbalfluencytasks,thetotalofnumbergenerated isthemostcommonlyusedmetric.

• However,onlythecorrectnumbersofitemsretrieveddoesnotprovideenoughinformationonunderlyingcognitivemechanisms involvedintheverbalfluencytasks(Troyer,2000).

• Inordertobetterunderstandthebehavioralprocessassociatedwiththeverbalfluencymeasures,researchersdevelopedadditionalmethodstoquantifyperformanceusingclusteringandswitchinganalyses(Troyer,Moscovitch,&Winocur,1997).

ClusteringandSwitchingAnalyses

Clustering• subcategorization oftheitemsthat

participantsgeneratedwithinaspecificcategory

• clusteringreliesonrelativelyautomaticprocessing byactivatingthesemanticrelatednesswithinacertainsubcategory.

Clusteringvs.Switching

Switching• Onceitemswithinasubcategoryare

exhausted,theyswitchtoanothersubcategory.Thisshiftingbehaviorwascalledasswitching(Troyeretal.,1998 )

• moreactivelyengagecognitiveprocessinordertoswitchthesubcategorytoanother

Chicken

Cow

Horse

Cat

Dog

Salmon

Trout

Farm

Pet

Fish

Cluster Switch

Clusteringvs.SwitchingDifferencesinneuralsubstrates

• Patientswithtemporallobelesions• Phonemic:notimpairedinboth

switchingandclustering• Semantic:smallercluster

• Participantswithfocalfrontallobelesions

• Lessswitchingthancontrols,butnormalclusteringperformance

ClusteringDisadvantageSwitchingDisadvantage

Bestpredictors• Phonemicswitching

• forfrontallesions• Semanticclustering

• fortemporallobelesions

Troyeretal.(1998).Clusteringandswitchingonverbalfluency:theeffectsoffocalfrontal- andtemporal-lobelesions.Neuropsychologia,36(6)

PurposeoftheStudy

Thecurrentstudyexamineddifferencesbetweenphonemic

andsemanticfluencyinADbyanalyzingclusteringand

switchinginage- andeducation-adjustednormativedata

(Troyer,2000).

Furthermore,weexaminedhowverbalfluencychangesover

timeandexploredwhatfactorspredictthesechanges.

Participants

Participantswere58individualswithprobableAlzheimer’sDisease(prAD)fromthedementiabankproject,PittCorpus(Beckeretal.,1994).

15participantswerefollowedupforasecondvisit(meandaysbetweenvisits:391.4,SD=47.2,Range=337-486)

1st Visit(n=58) 2nd Visit(n=15)Gender

(male:female)16:42 3:12

Age(yrs) 72.2(±8.8)(56-88) 1st Visit: 69.8 (±9.5) (56-88)2nd Visit:70.5(±9.5)(57-89)

MMSE 19.07(±4.04)(10-27) 1st Visit: 21.5 (±3.0) (16-27)2nd Visit:18.9(±4.6)(11-27)

Education(yrs) 11.7(±2.7)(6-20) 12.5(±2.6)(8-18)

Materials

Forsemanticfluency,participantsgeneratedanimalnames(1stvisit)andsupermarketitems(2ndvisit)for60seconds.

Forphonemicfluency,participantsgeneratedwordsbeginningwithf(1stvisit)ands(2ndvisit).

Semantic Fluency Phonemic Fluency

1st Visit Animal names Words beginning with F

2nd Visit Supermarket items Words beginning with S

Outcomemeasures

Dependentmeasuresincluded1) totalnumberofcorrectwords2) meanclustersize3) numberofswitches

- Clustersonsemanticfluencytrialsconsistofsuccessfullygeneratedwordsbelongingtothesamesubcategories.

SemanticFluencyAnalyses:Cluster(Troyer,2000)

Semanticfluencyanalyses(Troyer,2000)

Lion

Elephant

Zebra

Dog

Tiger

Jaguar

Lion

AfricanAnimal

Pet

Feline

#CR

1

1

1

1

1

1

1

0

Salmon Fish 1

#cluster #item/CL Clustersize

1

1

Cat

1AA

18 4

3

2

4

1

10

2

1

3

0

6Total(Sum)

SubcategoryExemplar

Dependentmeasures

Totalnumberofcorrectwords=

8

Meanclustersize=

Numberofswitches=

64

= 1.5

3

Semanticfluencyanalyses(Troyer,2000)

Phonemicfluencyanalyses(Troyer,2000)

- Clustersonphonemicfluencytrialsconsistofsuccessfullygeneratedwordsthatsharedanyofthefollowingphonemecharacteristics

PhonemicFluencyAnalyses:Cluster(Troyer,2000)

First Letters Words beginningwithsametwoletters

Arm,art

Rhymes Wordsthatrhyme Sand, stand

Firstandlast sounds Wordsdiffering onlybyavowelsound,regardlessoftheactualspelling

Sat,seat, soot,sight,sought

Phonemicfluencyanalyses(Troyer,2000)

Flake

Floss

Fun

Father

Fall

Fill

Wordsbeginningwith

sametwoletters

*Error!Wordswithdifferent

suffixes

#CR

1

1

1

0

#cluster #item/CL #CLsize

1

1

1

7 5

2

2

1

9

1

1

0

4Total(Sum)

SubcategoryExemplar

Feather

Funny

Ford

Wordsthatrhyme

*Error!Propernames

Wordswithdifferingonlybyavowelsound

1

1

0

1

1

1 2 1

1 2 1

Normalization

Therawscoreswerenormalizedbasedontheage- andeducation-adjustedcoefficients(Troyer,2000)

Therawscoresfor#CR(semantic)=35(age=50,Edu=13)Correctedscore=35+50*(0.23)+13*(-0.74)=36.8

Results- Semanticvs.Phonemiccomparisons(n=58)

One-wayRepeatedANOVAbetweenthesemanticandphonemicfluencymeasureswithcorrectedscorespereachDV

Semantic Phonemic

계열1 9.32 -4.96

-6 -4 -2 0

2

4

6

8

10

12

CORRECTEDSCORES

#Correctwords

Semantic Phonemic

계열1 0.59 0.49

0.59

0.49

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

CORR

ECTEDSCO

RES

MeanCluster Size

Semantic Phonemic

계열1 6.96 2.23

0

1

2

3

4

5

6

7

8

CORRECTEDSCORES

#Switching# CorrectWords MeanClusterSize #Switching

Lesswordsinphonemicthansemanticfluencytask,F (1, 57)=483.6, p

Results (Sem.Vs.Phon.N=58)

MMSE Age Education

#CorrectWords

Semantic 0.45** .096 -.264*

Phonemic 0.05 .095 -.643**

Mean Cluster Size

Semantic 0.02 -.027 -.205

Phonemic 0.26* .173 .101

#SwitchSemantic 0.32* .138 .017

Phonemic 0.13 .067 -.299*

Correlations&StepwiseRegression Analyses

Regression

MMSE&Educationp=.001

NONE

Educationp=.044

Results:Changesovertime

Mean SD One-wayANOVA

MMSE(1st visit) 21.47 3.04 Sig.F(1,14)=14.078,

p=0.002MMSE(2nd visit) 18.93 4.62

ResearchQ2:

weexaminedhowverbalfluencychangesovertimeandexploredwhatfactorspredictthesechanges.

Results– ModalityxTime-point

Two-wayRepeatedANOVAbetweenModality(Sem.Vs.Phon.)andTime-point(1st vs.2nd Visit)

Semantic Phonemic1stVisit 10 7.13

2ndVisit 9.27 7

0

2

4

6

8

10

12

correctedscores

# CorrectWords

Semantic Phonemic1stVisit 0.89 0.51

2ndVisit 1.07 0.43

0

0.2

0.4

0.6

0.8

1

1.2

Correctedscores

Semantic Phonemic1stVisit 6.4 5.53

2ndVisit 5.33 4.93

0

1

2

3

4

5

6

7

correctedsocres

MeanClusterSize #Switching

SignificantMainEffectsforModality:Phonemic<Semanticfor#correctwords,F(1,14)=2611.0,p

Results:Predictorforchangeovertime

Toexaminechangesovertime,firsttosecondvisitdifferencescoreswerecalculatedforeachdependentmeasure.

DV=Differencescoresbetween1st and2nd visitin#CR,Meanclustersize,and#Switchforsemanticandphonemicfluencytask

IVs=MMSE(1st visit),MMSE(2nd visit),Age,Education

MMSE(1st visit)scoressignificantlypredicteddiscrepancyscoresinnumberofswitchesforthephonemicfluencytask,F(1,13)=7.57,p=.016,R2=.368.

Nootherresultsweresignificant.

Discussion

ParticipantswithADdemonstratedlowerpercentilescoresinthesemanticthanphonemicclustering behaviors,consistentlywiththepreviousstudiesthatsuggestedADpatientswithtemporalatrophypresentedsemanticdisadvantage.

Semantic-ClusteringDisadvantage

Phonemic-SwitchingDisadvantage

IndividualswithADpresentedlessswitchingbehaviorsinphonemicthansemanticfluencytask.àPhonemic-switching behaviorshavebeenarguedasbeing

associatedwiththefrontallobefunctions.àWespeculatethatADparticipantsmayhavedeficitsin

frontallobe functionaswell.

Discussion

Peoplewithlowereducationdemonstratedgreaterdiscrepancybetweenphonemicandsemanticfluencytasks,giventhattheyearsofeducationwasthesignificantpredictortoaccountforthephonemic-semanticdiscrepancy.

Education hasbeenreportedasoneofthecriticalfactorsassociatedwithcognitivereserve inagingandADliterature.

Cognitivereserveisdefinedastheabilitiestooptimizeormaximizeperformancethroughdifferentialrecruitmentofbrainnetwork

Reducedcognitivereserveseemstoberesponsibleforgreaterdiscrepancybetweenthemodalities.

Educationascognitivereserve

Discussion

MMSEfromthe1st visitpredictedperformanceonchangesovertimeinphonemic-switchingbehaviors,whichisthemostfrontal-lobedemandingcondition.

à Itsuggestedthatcognitivereservefromthetimeofenrollmentmaybesensitivetodetectchangesovertimeespeciallyinthemostfrontal-lobedemandingbehaviors .

MMSE(1st Visit)&changesovertimeinphonemic-switching

ClinicalImplications:Phoneme-SwitchingTask:mayserveasanindexofcognitivereserve,

whichisimportantforcompensatingprogressivedeficits

Acknowledgement

• GlobalResearchNetworkgrantfromNationalResearchFoundationofKorea(NRF-2017S1A2A2038375)

• JeeEunSung (PI,Korea)• GayleDeDe (Co-I,USA)andSeunghun Lee(Co-I,Japan)