Sound-Symbol Learning inChildren with Dyslexia

Marjorie Gang and Linda S. Siegel

Abstract

This studcv evaluated the effect of sound-symbol association training on visual and phonological memory in children with a history ofdyslexia. Pretests of phonological and visual memory, a sound-symbol training procedure, and phonological and visual memoryposttests were administered to children with dyslexia, to children whose dyslexia had been compensated through remedial training, andto age- and reading level-matched comparison groups. Deficits in visual and phonological memory and memory for sound-symbol as-sociations were demonstrated in the dyslexia group. For children with dyslexia and children whose dyslexia had been remediated, thesound-symbol training scores were significantly associated with word and pseudoword reading scores and were significantly lower thanthose of the comparison groups. Children with dyslexia and children whose dyslexia had been compensated showed significantly lessfacilitation of phonological memory following the training than did typical readers. Skilled readers showed some reduction in accuracyof visual memory following the training, wvhich may be the result of interference of verbalization with a predominantly visual task. Aparallel decrease was not observed in the children wvith dyslexia, possibly because these children did not use the verbal cues. Childrenwith dvslexia and children whose dvslexia had been compensated seemed to have difficulty encoding the novel sounds in memory. Asa result, thev derived less phonological memory advantage and less visual memory interference from the training than did typical read-ers. Children in the compensated dyslexia group scored lower on sound-symbol training than their age peers. In other respects, the scoresof these children were equivalent to those of the typically reading comparison groups. Children in the compensated dyslexia group ex-hibited higher phonological rehearsal, iconic memorv, and associative memory scores than children in the dyslexia group. Implicationsfor the remediation of dyslexia are discussed.

C hildren with reading disabili-ties often experience extremedifficultv in learning the letter-

sound correspondences that constitutebasic decoding skills (Gillingham &Stillman, 1987; Siegel, 1986, 1993;Siegel & Faux, 1989; Snowling, 1980;Spector, 1995; Stanovich, 1986, 1988).Tests of decoding (i.e., reading of non-words) are considered among the bestdiagnostic measures of dyslexia (e.g.,Foorman, Francis, Fletcher, & Lynn,1996; Siegel & Heaven, 1986; Siegel &Ryan, 1988, 1989; Stanovich & Siegel,1994).

Many contemporarv researchers at-tribute the sound-svmbol associationdifficulties of individuals with dvs-lexia to phonological awareness (i.e.,perception) and encoding (i.e., mem-orv) limitations (Felton & Pepper, 1995;Foorman, Francis, Fletcher, Schat-schneider, & Mehta, 1998; Gathercole

& Baddeley, 1989, 1990a, 1990b; Mc-Dougall, Hulme, Ellis, & Monk, 1994;Torgesen et al., 1990; Torgesen et al.,2001; Vellutino & Scanlon, 1982; Wal-ton, Walton, & Felton, 2001). Verbalmemory span limitations in childrenwith dyslexia have been demonstratedon letter-string repetition and memoryscanning tasks (Farnham-Diggory &Gregg, 1975; O'Shaughnessy & Swan-son, 1998; Siegel & Ryan, 1988; Spring& Capps, 1974). These verbal memoryspan tasks are thought to reflectphonological encoding ability (Gather-cole & Baddeley, 1989, 1990a, 1990b;Gathercole et al., 1991; Hulme & Tord-off, 1989; Torgesen, 1988; Torgesen,Wagner, & Rashotte, 1994).

There is some empirical evidencethat part of the disability may be in vi-sual perception and memory for theforms of letters or their position in asequence (Badian, 1998; Cornoldi,

Rigoni, Tressoldi, & Vio, 1999; Eden,Stein, Wood, & Wood, 1995; Morrison,Giordani, & Nagy, 1977; Slaghuis,Lovegrove, & Davidson, 1993). Visuallimitations associated with dyslexiahave also been attributed to perceptual(Badcock & Lovegrove, 1981; DiLollo,Hansen, & McIntyre, 1983; Eden et al.,1995; Enns, Bryson, & Roes, 1995; Gal-aburda & Livingstone, 1993; Slaghuiset al., 1993) and memory processes(Corkin, 1974; Enns et al., 1995; Morri-son et al., 1977; Spring & Capps, 1974;Swanson, 1978, 1983, 1984).

The concept of visual perceptualdeficits in dyslexia has lately been con-sidered "thoroughly debunked" (Stano-vich, 1988, p. 601). However, there isevidence of an abnormality called visi-ble persistence in the early stages of pro-cessing of transient visual stimuli inindividuals with dyslexia. This percep-tual deficit could result in impaired vi-

IOI RNAL OF LEARNING DISABILITIES\OLL %IE 3., NU\IBER 2, MARCH/AI'RIL 2002, PAGES 137-1;7

JOURNAL OF LEARNING DISABILITIES

sual memory, especially for items insequences, and could interfere withreading by generating overlapping orsuperimposed images of letters (Di-Lollo et al., 1983; Eden et al., 1995; Far-mer & Klein, 1995; Klein & Farmer,1995; Slaghuis et al., 1993). Observa-tions about visible persistence andother studies of orthographic process-ing in individuals with dyslexia (Ba-dian, 1998; Cornoldi et al., 1999; Foor-man et al., 1996; Siegel, Share, & Geva,1995) have led to a resurgence of inter-est in visual memory and imagery asthey affect reading.

Visual memory deficits in childrenwith dyslexia have mainly beendemonstrated in studies of visual mem-ory span (Corkin, 1974; Morrison et al.,1977; Noelker & Schumsky, 1973; Senf& Freundl, 1971; Spring & Capps, 1974;Swanson, 1978,1983,1984). Some stud-ies of visual memory have shown noimpairment of immediate memory(Huba & Vellutino, 1990; Spring &Capps, 1974; Swanson, 1978, 1984,1986). For example, in two visual mem-ory studies, children with reading dis-abilities had equal or better memoryfor the last items in a sequence (iconicmemory). Their memory was only im-paired for earlier list items (Spring &Capps, 1974; Swanson, 1984).

The first hypothesis tested in thisstudy was that in comparison to chil-dren with typical reading skills, chil-dren with dyslexia have deficits in vi-sual memory for pseudoletters and inphonological memory for pseudo-words. Phonological and visual mem-ory span tests were used to assess thishypothesis. A test of immediate visual(iconic) memory was also adminis-tered. The second hypothesis tested inthis study was that scores on both vi-sual and phonological memory spantests would be positively correlatedwith reading scores.

In the experimental population weresome children who had a history ofdyslexia but who had developed read-ing skills above the 25th percentile as aresult of remedial reading instruction.There has been little research to dateconcerning children whose reading

difficulties have been compensated(Gallagher, Laxon, Armstrong, & Frith,1996). The presence of these children inthe sample made it possible to assessthe possibility that basic cognitivedeficits typically documented in read-ers with dyslexia might cease to exist ifremediation is successful.

Deficits in the integration of verbaland visual memory codes have beendemonstrated in children with a read-ing disability (Birch & Belmont, 1964;Ceci, Lea, & Ringstrom, 1980; Torge-sen, 1978-1979; Vellutino, Steger, Hard-ing, & Philips, 1975; see also Vellutino& Scanlon, 1982, for a review). Mem-ory studies using named and unnamedstimuli (Katz, Shankweiler, & Liber-man, 1981; Torgesen & Murphey, 1979;Vellutino & Scanlon, 1982) have indi-cated that skilled readers have a facil-ity for additive combination of visualand verbal cues relative to readers withdyslexia. The third hypothesis tested inthis study was that children withdyslexia have deficits in memory spanfor newly learned sound-symbol as-sociations in comparison with typicalreaders.

Deficits in phonological perceptionand memory (for letter sounds) and vi-sual perception and memory (for letterforms) may constrain the learning ofsound-symbol associations (Mauer &Kamhi, 1996). Associative memorydeficits in children with dyslexia may,therefore, be the direct result of verbaland visual deficits. Alternatively, defi-cits in associative memory may repre-sent a separate area of impairment, notcausally related to phonological or vi-sual deficits. To clarify this point, thefourth hypothesis tested in this studywas that visual and phonologicalmemory spans would be positivelycorrelated with memory span scoresfor newly learned sound-symbol cor-respondences in both typical readersand readers with dyslexia.

The hypothesis that learning thenames for symbols would increasememory span for the names and for thesymbols in typical readers but not inchildren with dyslexia was also tested.Swanson (1986) reported that sound-

symbol association training actuallyinterfered with recall in children withdyslexia. In a summary of several ofhis own studies dealing with verbaland visual coding processes in chil-dren with and without reading disabil-ities, Swanson (1986) concluded thatchildren of average intelligence whohave dyslexia fail to use multiple codesin an additive fashion in memorytasks. Swanson's studies showed thatlabel training increases memory for vi-sual forms in skilled readers but seemsto reduce recall in children with learn-ing disabilities (Swanson, 1978, 1983,1984). If children with dyslexia do notderive a memory benefit from usingmultiple codes, then perhaps thesound-symbol training aspect of phon-ics instruction only confuses them andinterferes with their reading skill de-velopment.

The sixth hypothesis tested in thisstudy was that in comparison to chil-dren with typical reading skills, chil-dren with dyslexia would show lowerrecall of items at the beginning of vi-sual and phonological sequences butsimilar recall of items at the end of se-quences. Difficulties with phonologicalencoding are often most profound foritems at the beginning of a sequence,representing a specific problem withphonological rehearsal (Baddeley, 1986).This is called a serial position effect. Se-rial position effects can demonstratestrengths and weaknesses in particularmemory strategies. Strong memory forbeginning list items (primacy) repre-sents the effective use of phonologicalrehearsal. Weak memory for theseitems suggests impaired rehearsal. Pro-ficient recall of end items (recency) mayreflect the duration of echoic memory,and weak recall of these items mayrepresent auditory confusion with ear-lier items (retroactive interference).

Method

Participants

Participants were volunteers from twoprivate schools. One school was Ken-neth Gordon School, a private elemen-

138

\ OlM\tE 35. \UMBER 2. MARCH/APRIL 2002

tarv school for students with severereading disabilities. The other schoolwas Pacific Academv, a Christian pri-vate school with elementarv and sec-ondarv divisions. Thirtv-one studentsfrom Kenneth Gordon School (ages 8vears 1 month to 13 vears 1 month)and 31 students from Pacific Academv(ages 6 years 9 months to 12 years 5months) w ere tested. The Kenneth Gor-don School stLdents all had histories ofsevere reading disabilities. Some of thestudents who had entered KennethGordon School with a history of read-ing disabilitv as defined in this study,after a vear or more of intensive reme-dial training, scored sufficientlv highon the Word Reading subtest of theWide Range Achievement Test-Thzird Edi-tion (WRAT-3; Wilkinson, 1993) andthe WA'oodcock Reading Mastery Test-Revised (WRMT-R; Woodcock, 1987)Word Attack subtest that they wvereclassified as the compensated dyslexiagroup (see Table 1).

The children w ith dyslexia hadWRAT-3 Word Reading subtest or

W\RMT-R Word Attack subtest pseudo-word reading scores at or below the26th percentile. The children whosedyslexia had been compensated hadWRAT-3 Word Reading subtest scoresand WRMT-R Word Attack subtestscores at or above the 26th percentilebut had entered the remedial languageprogram wvith measured reading scores

below that level. Comparison partici-pants had scores on both reading testsat or above the 35th percentile and nohistory of dyslexia. All participantshad IQ scores in the average range (80or higher) as indicated in their psy-choeducational test records or mea-sured by the Slossoni Intelligence Test(Slosson, 1981).

Each participant with dyslexia andeach participant whose dyslexia hadbeen compensated was assigned a spe-cific age and a specific reading levelmatch (matched for raw score on theWRAT-3) from among the typical read-ers. Thus, there were six groups:

1. children with dyslexia (D);2. tvpical readers who were age

matches for the children withdvslexia (D-A);

3. typical readers who were readinglevel matches for the children withdyslexia (D-RL);

4. children whose dyslexia had beencompensated (C);

5. typical readers who were agematches for the C group (C-A); and

6. typical readers who were readinglevel matches for the C group(C-RL; see Table 1).

Some students who were typical read-ers were used as matches in more thanone comparison group, resulting insome overlap of groups.

Experimental Tasks

For all participants, the phonologicalpretest was administered first, then theiconic memory pretest, and then the vi-sual memory span pretest. The sound-symbol association training followedthe pretests. Then the phonologicalposttest was administered, followedby the iconic memory posttest, andthen the visual memory span posttest.

Phonological Memory Test. Thetest of phonological memory (PM)span was based on a nonword repeti-tion measure described by Gathercoleand Adams (1993) and Gathercole andBaddeley (1989). One hundred twelvesingle-syllable, pronounceable pseu-dowords were used for the pretest, and10 of these pseudowords were used forthe posttest. The memory span taskconsisted of arrangements of pseudo-words into sequences of one, two,three, four, five, six, and seven ele-ments. The stimuli for the nonwordtests were recorded on a cassette tapeat a speed of 1 psuedoword per second.List lengths began at one and increasedto seven. There were four lists at eachlist length. At the end of each list, apause and a high tone signaled that thechild had up to 11 seconds to repro-duce the list. The posttest was similarto the pretest but consisted of se-quences constructed by random selec-

TABLE 1Demographic and Reading Skill Data on All Groups

Da D-Ab D-RLc Cd C-Ae C-RL'

Variable M SD M SD M SD M SD MD SD M SD

Age 10.73 1.27 10.70 1.26 8.11 1.48 10.76 1.27 10.46 1.16 8.81 1.75

Grade 4.58 1.62 4.50 1.24 2.08 1.24 5.21 1.44 4.26 1.15 2.79 1.48

WRAT-3Percentile 23.58 20.17 94.76 6.56 72.42 17.16 75.74 15.94 94.95 6.76 92.47 9.24Raw score 30.75 4.50 46.25 5.46 33.08 7.23 40.21 4.32 45.36 4.96 40.31 5.00

WRMT-RPercentile 24.58 14.93 82.17 16.62 72.42 17.16 67.79 15.13 81.89 16.86 81.81 17.78

Note. D = children with dyslexia: D-A = controls matched to D group by age: D-RL = controls matched to D group by reading level; C = children whose dyslexia hasbeen compensated: C-A = controls matched to C group by age: C-RL = controls matched to C group by reading level. WRAT-3 = Wide Range AchievementTest-Third Edition (Wilkinson. 1993): WRMT-R = Woodcock Reading Mastery Test-Revised (Woodcock, 1987).an = 12: 8 boys. 4 girls. bn = 12: 9 girls, 3 boys. In = 12: 10 girls. 2 boys. dn = 19; 11 girls, 8 boys. en = 19: 13 girls, 6 boys. fn = 19; 12 giris, 7 boys.

139

140 ~~~~~~~~~~~~JOURNAL OF LEARNING DISABILITIES

y1:yI

KmL0

0 U

UA-

U

U

T,K~Fmm U

FIGIon tl"LoctionLI-Tw~~Tillchoi1 Gae

ciati

0K

FIGURE 1. Pseudoletters used in the visual memory tests. Note. From "LocationConfusions in Visual Information Processing," by P. Dixon and L. Twilley, 1988,Canadian Journal of Psychology, 42, pp. 378-394. Copyright 1988 by CanadianPsychological Association. Adapted with permission.

tion with replacement from the 10pseudowords used in the training in-tervention, so there were many fewerto remember.

Iconic Memory Test. This test usedgraphic displays on a computer moni-tor to assess iconic visual memory andreaction time. Thirty nonmeaningfulsimple line drawings (pseudoletters;Dixon & Twilley, 1988) were used as vi-sual stimuli in this test (see Figure 1).Two-by-two grids of these pseudolet-ters were shown on a computer screenfor 2 seconds (see Figure 2). A Macln-tosh LC III computer and VScope soft-

ware were used. Following presen-tation of each grid, a new grid was pre-sented on the computer screen thatcontained only a single probe characterin one of the grid positions. The otherboxes in the grid were empty. The par-ticipant's task was to determine whetherthe probe character was contained inthe same position as in the previousgrid. The participant pressed keys onthe computer keypad marked there andnot there to indicate presence or ab-sence of the target characters. The tri-als were presented in the same randomorder for each participant. The child'schoices and the correct choices were

IURE 2. A sample array as shownhe computer screen. Note. From~ation Confusions in Visual Informa-Processing," by P. Dixon and L.

ley, 1988, Canadian Journal of Psy-logy, 42, pp. 378-394. Copyright3 by Canadian Psychological Asso-on. Adapted with permission.

recorded by the Vscope software in aninternal data file. Latency scores werealso recorded by the computer in thisdata file.

For the posttest, the four-elementgrids were constructed by selectionfrom the 10 pseudoletters selected forthe posttest and used in the training in-tervention (see Figure 3). Pre- andposttesting were conducted in blocksof 20 trials.

Visual Memory Span Test. The vi-sual memory span task was based onone used by Katz, Shankweiler, andLiberman (1981). The 30 nonmeaning-ful simple line drawings used for theiconic memory test were used in thistest as well. This test was also pre-sented on a Maclntosh LC III compu-ter, using Vscope software. Pseudolet-ters were arranged into sequences oftwo, four, and six elements. Elementsin the sequences were paired so thattwo pseudoletters appeared at a time,each pair appearing in the center of thescreen at a rate of about I pair per sec-ond. Each trial consisted of a sequencefollowed by a colored probe character.The participant's task was to deter-mine whether the probe character had

N4-K-2

T

f

m0 -F

140

VOLUME 35, NUMBER 2, MARCH/APRIL 2002

been contained in the sequence. Eachparticipant completed a block of 20 tri-als, including 5 two-element lists, 10four-element lists, and 5 six-elementlists for both the pretest and theposttest. For the posttest, the se-quences were constructed bv selectionfrom the 10 pseudoletters selected forthe training procedure and the post-test. Posttesting was also conducted inblocks of 20 trials.

To calculate serial position scores,only sequences of four and six visualelements were used. The number ofcorrect responses to probes corre-sponding to pseudoletters in the be-ginning (first pair), middle (secondpair), and end (last pair) of the se-quence were recorded. These scoreswere then converted according to thefollowing equation:

number of correct responsesmaximum possible correct

10 = serial position score

This allowed the recording of fivedependent measures for each temporaltest: total number correct, number ofprimacv items (i.e., first shown stimuliin each sequence) correct, number ofmiddle items correct, number of re-cencv items (i.e., last shown stimuli ineach sequence) correct, and responselatencv.

Training Procedure

A training procedure followed thethree tasks. Ten of the pseudolettersand 10 of the pseudowords were se-lected for this test. Children weretaught to associate each visual elementwith a specific phonological element;that is, the invented letters used in thevisual memory span test were assignednames from the list of nonwords usedin the nonword repetition test (see Fig-ure 3). Items were presented in thesame order for all participants.

Pictures of shapes on cards were pre-sented and names supplied in iterativecycles either until the participant wasable to name all of the stimuli in twocomplete presentations of the deck of10 svmbols or until the entire deck hadbeen presented 20 times. The experi-

menter presented the stimuli on 3 x 5inch index cards, in the first two pre-sentations supplying the name ("Thisone is called ... "). After the first twopresentations, the experimenter waiteda few seconds to allow the child to sup-ply the name, then requested the nameof the stimulus ("What sound goeswith this shape?"). The experimenterprovided the name if the child wasunable to do so after 5 seconds ("Thisis . . .")

The number of cycles necessary toachieve the criterion level of masteryor the number of names mastered in 20repetitions of the deck was recordedon the phonological memory codingsheets. The training score recordedrepresents the average number ofnames leamed per one repetition of thedeck.

prin

jix

lish

tweg

/Nl

Results

Question 1Do children with a history of dyslexiahave deficits in phonological or visualmemory relative to chronological andreading age-matched children withtypical reading skills?

Phonological Memory. To test thehvpothesis that children with readingdisabilities have deficits in phonologi-cal or visual memory relative to chron-ological and reading age-matched chil-dren with typical reading skills, thescores of D, C, and control group chil-dren on visual and phonological taskswere compared. The mean scores ofD, C, and their respective comparisongroups on the tasks are shown in Ta-ble 2.

Group mean pretest and posttestscores on memory measures for groupsD and C were compared to the groupmean scores of the D-A and D-RL andC-A and C-RL comparison groups, re-spectively, with a repeated measuresanalysis of variance (ANOVA). The3 x 2 repeated measures analyses thatwere conducted represent three read-ing groups and two times of testing(pre- and posttest). Six of the 3 x 2 com-parisons were computed, comparing D

larp YFIGURE 3. Symbols used in the train-ing intervention and posttests. Note.From "Location Confusions in Visual In-formation Processing," by P. Dixon andL. Twilley, 1988, Canadian Journal ofPsychology, 42, pp. 378-394. Copyright1988 by Canadian Psychological Asso-ciation. Adapted with permission.

to D-A and D-RL and comparing C toC-A and C-RL for the phonologicaltests, the visual iconic memory tests,and the visual temporal memory tests.Post hoc Tukey HSD tests were usedfor the three group comparisons. Themean memory scores of group D werealso compared to those of group C by3 x 2 repeated measures ANOVA of Dand C groups, and all matched typicalreaders participating in the study (seeTable 2).

141

142

-.- -) O N _ O N N -0

t e7 1< > NO OD D O 2-)N- 1 0) COtO O) o

C) U0 .) C- _ _ C

C.) -- t LO N N iC r N NO CON 0) 0

co) C - CO

tl) _ NO) 0 0

CJa 0C CM N _ C\N CO UC\l r~~~- co

> t On On Ce ~~~~~~~~~~O (Uo

_ U)

.0 ' COD L_ CM c OCO 0 ENl-IC) C)COi COl) U? 0

(U CO)C\J CO)CO) aC 0

co C\ - C .0 -~~~0a 0 - Co

O ~ ~~~~ C)U h O~ rl D 'P 7 D Ol C

0 NU N a,0 = c ><0) CON- CO.- 0 E (U

ffi ~ -C CO 0 nt N CO- CO OO

E Cl _ NN8 co

E t 2

0) ~~~~~~~~~ ~4) C

U7 CD a) a) _ Qo ce c: i-a

X ~ Cco \ (o .C 00

a0 a) NIL N.t- CO) oC

> cq -n tt S2s

O~~~~~O O 0 N co -l t ne S

L.NO co U) C a U) (D u)CO 1-: -6 C~ C; C8 s )

o C CO A

~0 - - C

0 ° o ° E o o c 0 S°C 8 > \ LO

CL C6 .-L0 CO NCo N 0~ o

0 c(Ot .0 NN O

o4 CO

CU 0

o ~ CO 1* -C\ CO\O CO -

C OLt NOLO LON LO L E U

co co c CO C\l ItCO 0

C[ 0 coC

LLIo ~~~~~o 0C CO co cc 20 cn~~~~~~~~~~~~~

C0 ClC)O co -LOL 0 - to V

-J C o pLA .9r

CYNC NN C NN C la

*0 &~~~c 0 c

(U~~~~C COO 0). LO- COD) C\O 2 0 .-- -N Nl .0-C

0 a)

~~LLL) COOc -r NO C ) C\J NN LO L t COLt V T

o - 02~~~~~~~~~~~~~- EO

c ~ ~ ~ ~ ~ ~ ~ ~ ~ a~~~~0

CON N- ~~~~~u)uNN 1

N-) O N 0 CO Er>

o COCO NNE-CO O o CCOLt) CO)CCD

c'o co~ NC 00 II(UN-CDE ) E- - a . 0 >-6, (D a) r- 0~~C (UC

o t Z , E cD U) I) Io-

. EL (L L ) EL

0 co~~~~

-~OO - o2

a.> t

COLUMIE 35. NUMBER 2, MARCH/APRIL 2002

On the phonological measures, re-peated measures 3 x 2 ANOVA com-parison of phonological pretest andposttest scores for group D with D-Aand D-RL showed a significant maineffect for group, F(1, 33) = 6.22, p = .005,ES = .27. The post hoc Tukey HSD testshowed that the mean phonologicalpretest score of the D group was notsignificantly lower than the corre-sponding scores for either the D-A orthe D-RL groups (see Table 2). How-ever, relative to the D-A group, Dgroup readers scored significantlylower on the phonological posttest, p =.001 (see Table 2). Therefore, in this in-vestigation children in the D group diddemonstrate significant deficits inphonological memory relative to D-Achildren on the phonological posttest,although not on the phonologicalpretest scores.

In comparison to vounger studentsmatched for reading level, D groupstudents did not have significantlylower mean scores on either thephonological pretest or the phonologi-cal posttest (see Table 2). There were,however, significant differences be-tween the D group and the D-RL groupon all of the phonological posttest ser-ial position variables in a 3 x 2 repeatedmeasures comparison (see Table 3). Arepeated measures 3 x 2 ANOVA ofphonological pretest and posttest pri-macy scores showed a significant maineffect for group, F(2, 33) = 3.38, p = .046,

and a significant interaction effect,F(2, 33) = 5.17, p= .011, ES = .239.D group children showed a significantdeficit in phonological posttest pri-macy scores representing pseudo-words in the beginning of posttestmemory lists in comparison to theirage peers, p = .05, and to their readinglevel peers, p = .002, according toTukey post hoc analysis (see Table 3).Scores for pretest primacy did not dif-fer significantly among these groups(see Table 4).

Similarly, a repeated measuresANOVA of phonological pretest andposttest middle position scores showeda significant main effect for group, F(2,33) = 4.55, p = .016, and a significant in-teraction effect, F(2, 33) = 4.40, p = .02.Children with dyslexia showed a sig-nificant deficit in phonological post-test middle position scores in compar-ison to their age peers, p = .005, andtheir reading level peers, p = .04, ac-cording to Tukey post hoc analysis (seeTable 3).

Repeated measures ANOVA ofphonological pretest and posttest re-cency scores showed a significant in-teraction effect, F(2, 33) = 4.66, p = .016,but no significant group main effect. Inthe C group and in all control groups,the recency effect that was evident onthe pretest was reduced or eliminatedon the posttest. This did not occur inthe D group. Children in this groupshowed a significant recency effect on

both pretest and posttest (i.e., signifi-cant elevation of recency score relativeto middle position score). Childrenwith dyslexia showed a significantdeficit in phonological posttest recencyscores in comparison to their age peers,p = .032, and their reading-level peers,p = .003, according to Tukey post hocanalysis (see Table 3), ES = .22 for theinteraction. Thus, children in the Dgroup did show significant deficits inphonological memory relative to D-RLreaders with respect to individual ser-ial position scores but not on thephonological pretest and posttest totalscores.

A repeated measures 3 x 2 ANOVAcomparison of phonological pretestand posttest scores for the C groupchildren with C-A and C-RL groupsdid not show a significant main effectfor group (see Table 2). Furthermore,the C group did not show significantdeficits at any of the phonologicalserial positions (see Tables 3 and 4).Children in the C group did notdemonstrate significant deficits inphonological memory relative to C-Aor C-RL groups.

Visual Memory. Although the Dgroup had lower mean scores on all ofthe visual tests than either of the com-parison groups, significant differencesin mean visual memory scores wereevident only between D group chil-dren and D-A group children on the

TABLE 3Phonological Posttest Serial Position Scores by Groups

Primacy Middle Recency

Group M SD pa M SD pa M SD pa

D .71 .25 .55 .33 .60 .27

D-A .89 .16 .054 .87 .18 .007 .80 .13 .029

D-RL .98 .05 .004 .79 .16 .033 .87 .12 .006

C .94 .10 .74 .22 .72 .23

C-A .92 .13 ns .80 .20 ns .81 .16 ns

C-RL .89 .16 ns .73 .24 ns .79 .17 ns

Note. D = children with dyslexia; D-A = controls matched to D group by age; D-RL = controls matched to D group by reading level; C = children whose dyslexia hasbeen compensated; C-A = controls matched to C group by age: C-RL = controls matched to C group by reading level.aSignificance scores for t test comparsons between disability groups and matched control groups.

143

144 JOURNAL OF LEARNING DtSABtLITIES

iconic memory pretest (see Table 2).For children in the D group and D-Aand D-RL groups, a repeated measuresANOVA of iconic memory pretest andposttest scores showed a significantmain effect for group, F(2, 35) = 4.37,p = .021, ES = .21. According to Tukey

post hoc analysis, D group readersscored significantly lower than theirD-A peers on the iconic memory pre-test, p = .021. The visual memory spanpretest and posttest 3 x 2 analysis didnot show a significant main effect forreading group.

A multivariate 3 x 2 ANOVA for theD group and both of its comparisongroups also showed significant differ-ences between the D group and theD-A group on the iconic memorypretest, F(2, 34) = 3.47, p = .035, and thevisual memory span pretest, F(2, 34) =3.28, p = .049, according to post hocTukey analysis. In summary, childrenwith reading disability demonstratedvisual iconic memory deficits relativeto age peers with typical reading skillsbut not relative to younger typicalreaders matched for reading level.They did not show statistically signifi-cant deficits in visual memory spanrelative to either comparison group.

There was a significant main groupeffect on mean visual response latency,F(2, 35) = 3.32, p = .048. However, posthoc Tukey analysis did not show anysignificant two-group differences. Youn-ger children were generally slower torespond manually to the visual probesthan older ones, regardless of readingability. Children with dyslexia scoredsimilar to their age peers on visual re-sponse latency.

Children in the C group did notdemonstrate significant iconic mem-ory deficits relative to either of the typ-ically reading comparison groups (ageor reading level matches). The C groupdid not score significantly lower thaneither C-A or C-RL groups on thevisual memory span tests. There wasa significant difference between theC group and the C-RL group on vi-sual response latency, F(2, 56) = 4.45,p = .016, ES = .17. The mean visual re-sponse speed of the older children was

higher than that of the younger read-ing level-matched group.

In some of the comparisons, visualmemory scores were significantly cor-related with age. In the typical read-ers, the iconic memory pretest scorewas significantly associated with age,r(30) = .65, p < .001, as was the iconicmemory posttest score, r(30) = .41, p =.025, and the visual memory spanpretest score, r(30) = .38, p = .039. For

the C group children, the iconic mem-ory pretest was significantly associatedwith age, r(19) = .58, p = .009. Therewas significant correlation of age andvisual response latency, reflecting thefact that the youngest children wereslower to respond to the computer. Theanalysis of covariance (ANCOVA) pro-cedure was therefore used for theanalysis of visual memory, using age asa covariate. This increased the level ofsignificance of group differences in thegroups that were matched by age. Forthe D group students and matchedD-RL students, the use of the covariateanalysis resulted in some group meandifferences that were significant at the5% level. Significant differences werenoted for the iconic memory pretest,F(1, 23) = 5.15, p = .034, and the iconicmemory posttest, F(1, 23) = 4.65, p =.043. D group children scored signifi-cantly lower than the D-RL compar-ison group on both iconic memorymeasures. Phonological memory span

scores were not significantly associ-ated with age in the groups tested.

The mean memory scores of theD group readers were compared tothose of C group readers by 3 x 2 re-peated measures ANOVAs of D andC groups and all typical readers par-ticipating in the study (see Table 2).C group children had higher meanscores than D group children on all thevisual memory tests, but only on theiconic memory pretest was the differ-ence significant. A repeated measuresANOVA showed a significant groupeffect, F(2, 58) = 3.80, p = .028, ES = .12.The score of the D group was signifi-cantly lower than that of the C group,p = .014, by a Tukey test.

The D group also had a lower meanscore on the phonological memoryposttest than the C group. The main ef-fect for group was F(2, 58) = 3.11, p =.052. According to post hoc Tukeyanalysis, the D and C groups were sig-nificantly different, p = .039. The C groupscored significantly higher than theD group on phonological primacy onboth the pretest and the posttest, F(2,58) = 7.33, p = .023 and p = .001, re-

spectively, ES = .20.

Question 2

Are scores on the measures of pho-nological or visual memory signifi-cantly associated with reading skills in

TABLE 4Phonological Pretest Serial Position Scores by Groups

Primacy Middle Recency

Group M SD M SD M SD

D .52 .39 .45 .33 .64 .36

D-A .81 .16 .65 .17 .69 .15

D-RL .57 .35 .40 .27 .52 .33

C .79 .17 .59 .15 .71 .19

C-A .79 .15 .64 .17 .72 .19

C-RL .76 .17 .55 .18 .64 .19

Note. All t test ccomparisons between disability groups and matched control groups were nonsignificant.D = children with dyslexia; D-A = controls matched to D group by age; D-RL = controls matched to D groupby reading level; C = children whose dyslexia has been compensated; C-A = controls matched to C groupby age; C-RL = controls matched to C group by reading level.

JOURNAL OF LEARNING DISABILITIES144

VOLUME 35, NUMBER 2, \IARCi/ APRIL 2002

children with a historv of dyslexia orin chronologically and reading age-matched children with tvpical readingskills? To determine whether measuresof phonological or visual memorywere significantlv associated with

reading skills in children in the D andC groups or in the control groups, par-tial correlations (controlling for age)were employed. To obtain a score forall control participants, data from allmatched comparison groups were

TABLE 5Significant Correlations Between Memory Variables and Reading

Scores for All Participants (N = 60)

WRAT-3 WRMT-R

Memory variable r p r p

Phonological memoryPretest

Total .28 .033 .36 .005Primacy .42 .001 .42 .001Middle .36 .005 .38 .003Recency ns ns

PosttestTotal .42 .001 .38 .003Primacy .34 .007 .33 .009Middle .35 .006 .38 .003Recency ns ns

Iconic memoryPretest .37 .003 ns

Sound-symbol training score -.60 .001 .66 .001

Note. WRAT-3 = Wide Range Achievement Test-Third Editiorn WRMT-R = Woodcock Reading MasteryTest-Revised. Participants' scores on WRAT-3 and WRMT-R were significantly correlated, r = .87, p = .001.All visual memory span tests were nonsignificant, with the exception of the posttest prmacy score, whichwas significantly correlated with the WRAT-3 (r = .33. p = .01).

TABLE 6Significant Correlations Between Memory Variables and Reading

Scores for Matched Groups of Typical Readers

WRAT-3 WRMT-R

Memory variable r p r p

Phonological memoryPretest

Total .37 .050 .44 .018Primacy .57 .001 .46 .012Middle .55 .002 .46 .013Recency .63 .001 .58 .001

Visual memory spanPosttest

Primacy .48 .008 nsMiddle .44 .017 .41 .029Recency ns ns

Sound-symbol training score -.36 .055 .41 .026

Note. WRAT-3 = Wide Range Achievement Test-Third Edition; WRMT-R = Woodcock Reading MasteryTest-Revised. Participants' scores on WRAT-3 and WRMT-R were significantly correlated, r= .79, p = .001.All phonological posttest correlations were nonsignificant. All visual memory span pretest correlations werenonsignificant.

combined. Mean scores on the visualand phonological measures were com-pared to mean scores on the readingtests. Significant correlations betweensome of the measures and the readingscores were obtained, as listed inTables 5 and 6. When all participants inthe study were considered as a group,phonological pretest and postest scoresand serial position scores for items inthe primacy and middle positions ofmemory lists were significantly associ-ated with both word and pseudowordreading. Considering only the typicalreaders, all phonological pretest serialposition scores were significantly asso-ciated with word reading and pseudo-word reading (see Tables 5 and 6).

Question 3

Do children with a history of dyslexiahave deficits in memory span forletter-sound correspondences relativeto chronologically and reading age-matched children with typical readingskills? ANOVAs of sound-symboltraining scores were used to determinewhether children with dyslexia had def-icits in memory span for letter-soundcorrespondences relative to chronolog-ically and reading age-matched chil-dren with typical reading skills. Thescores for the sound-symbol associa-tion task were presumed to representmemory span for sound-symbol corre-spondences (see Table 7).

A three-group ANOVA showed asignificant main effect for group, F(2,35) = 8.76, p = .001, ES = .53. D groupparticipants had significantly lowerscores on training than either D-A chil-dren, p = .001, or D-RL children, p =.025, by post hoc Tukey analysis. Chil-dren with reading disability haddeficits in memory span for novelletter-sound correspondences relativeto chronologically and reading age-matched children with typical readingskills.

An ANOVA of sound-symbol train-ing scores of the C group and its com-parison groups showed a significantmain effect for group, F(2, 56) = 5.04,p = .01, ES = .19. The C group's train-

145

146 JOURNAL OF LEARNING DISABILITIES

TABLE 7Comparison of Dyslexia Groups and Matched Controls on Memory Scores for Letter-Sound Correspondences

D D-A D-RL C C-A C-RL

Variable M SD M SD M SD M SD MD SD M SD

PretestPrimacy 5.27 2.41 6.83 1.80 6.83 2.17 6.00 2.83 7.16 1.80 7.37 2.00Recency 9.09 3.02 9.16 1.95 8.33 2.46 8.33 2.43 9.21 1.87 9.21 1.87

PosttestPrimacy 4.77 2.36 5.83 3.26 4.79 3.10 5.92 2.66 6.05 2.68 5.79 2.89Middle 6.36 5.04 9.17 2.89 5.83 5.15 5.79 5.06 6.84 4.78 7.37 4.52Recency 8.18 3.37 9.17 1.95 7.08 3.34 7.89 3.03 8.16 2.99 8.16 2.99

Training .43 .26 .93 .29 .77 .34 .66 .22 1.01 .39 .89 .39

Note. D = children with dyslexia; D-A = controls matched to D group by age; D-RL = controls matched to D group by reading level; C = children whose dyslexia hasbeen compensated; C-A = controls matched to C group by age; C-RL = controls matched to C group by reading level.

ing score was significantly lower thanthat of the C-A group by Tukey test,p = .008. C group children had deficitsin memory span for novel letter-soundcorrespondences relative to C-A matchedchildren. The C group did not scoresignificantly lower than the C-RLgroup.

Participants whose dyslexia hadbeen compensated did score signifi-cantly lower on sound-symbol train-ing than a combined group of all thetypical readers in the study, F(2, 60) =10.54, p < .001, ES = .36. The meantraining score for the C group was notsignificantly different from that of theD group in a 3 x 2 ANOVA. However,in a two-group comparison of D andC group participants, there was a sig-nificant difference in mean trainingscores when age was used as a covari-ate, F(1, 37) = 7.52, p =01.

Question 4

In children with a history of dyslexia,are memory span scores for sound-symbol correspondences related to vi-sual and phonological memory scores?To determine whether memory spansfor letter-sound correspondences weresignificantly related to visual andphonological memory scores, partialcorrelations (controlling for age) wereused. Partial correlations between thephonological measures and sound-

symbol training scores were not signif-icant in the D group or in typical read-ing groups. Training scores were alsonot significantly associated with any ofthe visual measures in the D group. Fortypically reading students in Grade 1(i.e., the D-RL group), training scoreswere significantly associated withphonological posttest primacy, r(I0) =.85, p = .002, and phonological posttestrecency, r(l0) = .79, p = .006. Phonolog-ical memory may play a limiting role inmemory for sound-symbol correspon-dences in very young but not in olderchildren.

Question 5

Do learned sound-symbol correspon-dences differentially affect visual andphonological memory for childrenwith a history of dyslexia and theirpeers? To investigate the hypothesisthat learned sound-symbol associa-tions differentially influence visualand phonological memory for childrenwith a history of dyslexia and theirmatched peers, the differential effectsof the sound-symbol training were an-alyzed by a 3 x 2 repeated measuresANOVA. The three levels of group rep-resented comparisons of the D groupto each control group and then of theC group to each control group. The twolevels of test were the pretest and theposttest. Interaction effects were ana-

lyzed by repeated measures ANOVAand ANCOVA using age as a covariate.

All groups had higher mean phono-logical posttest scores than meanphonological pretest scores. The maineffects for pretest-posttest were signif-icant for D and comparison groups,F(1, 33) = 88.02, p < .001, ES = .73, andfor C and comparison groups, F(1, 54)= 91.33, p < .001, ES = .63. The phono-logical posttest was presumably easierfor the children than the pretest be-cause the pseudowords were familiarand because there were fewer of themin the posttests (only 10 as opposed to112 for the pretest). There may alsohave been a facilitative effect of multi-ple coding among typical readers be-cause phonological posttest scores aresignificantly associated with sound-symbol association scores for thisgroup. The association of phonologicalposttest and training scores was notsignificant in the D group. Significantinteraction effects were obtained forthe phonological tests in comparisonsinvolving the D group, F(1, 33) = 6.22,p = .005. The mean score increase frompretest to posttest was significantlyless for the dyslexia group than for thematched comparison groups. Phono-logical pretest-posttest interaction ef-fects were not significant in compar-isons involving the C group.

Following the sound-symbol train-ing procedure, mean iconic memory

VOLUME 3~~, NUMBER 2, MARCH/APRIL 2~~2 147

scores decreased. The main pretest-posttest effects for iconic memory weresignificant by 3 x 2 repeated measuresANOVA for D and comparison groups,F(1, 33) = 4.68, p = .038, and for C andcomparison groups, F(1, 54) = 13.10,p = .001, ES = .20. Mean iconic memoryposttest scores were lower than iconicmemory pretest scores for all groups.The mean pretest-posttest differenceobserved in the D group alone was notsignificant when the scores were com-pared by ANOVA, ANCOVA, or t test.Following the sound-symbol associa-tion training, an ANOVA for visualmemory span pretest and posttestscores did not show a significant meanscore change for the D group and itscomparison groups or for the C groupand its comparison groups. There wereno significant interaction effects foriconic memory or visual memory span.

primacy

Question 6

Are serial position curves for visualand phonological memory spans sig-nificantlv different for children with ahistory of dvslexia and their typicallyreading peers? To determine whetherserial position curves for visual andphonological memory spans were sig-nificantlv different for children withdvslexia and their typically readingpeers, mean serial position scores forthe D and C groups and a third groupcomposed of all typically reading par-ticipants were compared. To obtainmean phonological serial positionscores, a count of correct phonemes forbeginning, middle, and end positionsin the four-pseudoword lists wasrecorded for each child. The four-element lists were chosen because allbut two of the participants receivedsome score for this list length, but onlytwo had no errors at this length.Phonological serial position curves aretypically U-shaped, higher at the be-ginning and at the end and depressedin the middle. Phonological primacyscores theoretically reflect the use ofcumulative rehearsal of beginning andmiddle list items. Recency scores rep-resent the superioritv of immediate

middle

Serbil poston

recency

* pres

------ = dyeda

__ = compensatsd

=coriro

FIGURE 4. Phonological pretest and posttest serial position scores for all groups.

auditory memory in the absence of in-terference. All groups showed phono-logical pretest serial position effects(see Table 4 and Figure 4), with somevariations.

A repeated measures 3 x 2 ANOVAof phonological pretest and posttestprimacy scores showed a significantmain effect for group, F(2, 33) = 3.38,p = .046, and a significant interactioneffect, F(2, 33) = 5.17, p = .011. D groupchildren showed a significant deficit inphonological posttest primacy scores,representing pseudowords at the be-ginning of posttest memory lists, incomparison to both their age peers, p =.05, and their reading-level peers, p =.002, according to Tukey post hocanalysis (see Table 6). Scores for pretestprimacy did not differ significantlyamong these groups.

Similarly, repeated measures ANOVAof phonological pretest and posttestmiddle position scores showed a sig-nificant main effect for group, F(2,33) = 4.55, p = .018, and a significant in-teraction effect, F(2, 33) = 4.40, p = .02.Children with dyslexia showed a sig-nificant deficit in phonological posttestmiddle position scores in comparisonto both their age peers, p = .005, andtheir reading-level peers, p = .04, ac-cording to Tukey post hoc analysis (aspreviously noted; see Table 3).

Repeated measures ANOVA ofphonological pretest and posttest re-cency scores showed a significant in-teraction effect, F(2, 33) = 4.66, p = .016,but no significant group main effect. Inthe C group and the typical readinggroups, the recency effect that was ev-ident on the pretest was reduced or

1.00

Phofcgimlscore 0.95

0.90

0.85

0.80

0.75

0.70

0.85

0.80

0.55

0.50

0.45

0.40

NR

a---------

147VOLUME 3;, NUMBER 2, MARCH/APRIL 2002

JOURNAL OF LEARNING DISABILITIES

eliminated on the posttest. This did notoccur in the D group (see Tables 3 and4). Children with dyslexia showed asignificant deficit in phonologicalposttest recency scores in comparisonto both their age peers, p = .032, andtheir reading level peers, p = .003, ac-cording to Tukey post hoc analysis (aspreviously noted; see Table 3). C groupchildren scored similarly to C-A andC-RL group children at all phonologi-cal serial positions. Phonological serialposition curves of C group children,therefore, were not significantly differ-ent from those of C-A or C-RL groupreaders.

A repeated measures 3 x 2 ANOVAof phonological primacy scores of D andC groups and a combined group of alltypically reading participants showeda significant main effect for readinggroup, F(2, 58) = 7.33, p = .001. TheD group scored significantly lowerthan the C group on phonologicalpretest primacy, p = .023, and posttestprimacy, p = .001, by post hoc Tukeyanalysis.

Phonological primacy scores weresignificantly higher on the posttestthan the pretest for all comparisons.Most typical and compensated readersreceived their highest mean scores foritems at the beginning of the lists. In acomparison of D, C, and all typicalreading groups, a repeated measures3 x 2 ANOVA showed a significantphonological primacy score increase(i.e., significant main effect for testtime) from pretest to posttest, F(1, 58) =28.26, p < .001, ES = .33, with no signif-icant interaction. Phonological middleposition scores were also higher on theposttest than on the pretest, F(1, 33) =36.30, p < .001. In this case, the inter-action was significant. The pretest-to-posttest gain at this position was lessfor children with dyslexia than for theother groups, suggesting possible en-coding or rehearsal deficits.

There was a reduced but still evidentrecency effect in the dyslexia group onthe phonological posttest, in contrastto the phonological pretest. For theC group and the typical readers, there

was no recency effect on the phonolog-ical posttest, although there was oneon the pretest, F(1, 58) = 22.86, p < .001,with no significant Group x Positioninteraction.

Among typical readers, all phono-logical pretest serial position scoreswere significantly correlated with bothword and pseudoword reading scores.Among D group readers, no significantcorrelations between phonologicalpretest serial position scores and read-ing scores were obtained.

Visual serial position curves are gen-erally different in form from phono-logical curves (see Figure 5). They donot show a primacy effect (see Table 7),suggesting that visual cumulative re-hearsal does not occur. Visual memoryspan primacy scores were significantlyhigher on the pretest than on theposttest for some of the group compar-isons. Repeated measures ANOVA forthe D group and its comparison groupsshowed a significant main effect fortest time, F(1, 31) = 4.53, p = .041. For theC group and its comparison groups,there was also a significant main effectof test time, F(1, 53) = 4.80, p = .033. Vi-sual memory span recency scores weresignificantly lower on the posttest thanon the pretest for C and its comparisongroups, F(1, 53) = 5.80, p = .02, but notfor D and its comparison groups.

The visual memory span posttestprimacy score and posttest middle po-sition score, representing visual recallof a newly learned visual character withvisual retroactive interference, were sig-nificantly correlated with WRAT-3 wordreading scores in typical readers, r(27)= 48, p = .008, and r(27) = .44, p = .017,respectively. The visual memory spanposttest primacy score was also signif-icantly correlated with word reading inthe entire sample, r(58) = .33, p = .010.Visual memory span posttest middleposition scores were significantly cor-related with pseudoword reading inthe children with typical reading skills,r(27) = .41, p = .029. None of the visualserial position scores were signifi-cantly correlated with reading scores inthe D group.

Discussion

Phonological Memory

With respect to phonological memory,it was expected that in comparison tochildren with typical reading skills,children with dyslexia would havedeficits in phonological memory fornovel names. Reading ability-relateddifferences in phonological memorywere demonstrated in this study. Chil-dren with dyslexia showed deficits inphonological memory, as indicated bylower posttest scores and lower scoresat all posttest serial positions thaneither comparison group. Significantassociations between word and pseu-doword reading scores and the phono-logical measures, especially phono-logical primacy, were obtained in thetypically reading group. The onlyphonological score significantly asso-ciated with reading in the dyslexiagroup was the primacy score, possiblyreflecting a deficit in phonological en-coding or rehearsal that may limitreading ability in this group. Childrenwhose dyslexia had been compensateddid not have significantly lower meanscores at any of the phonological serialpositions than the comparison groups.

Comparing the mean phonologicalpretest scores of children in the D groupto those of their age- and reading-level-matched peers produced no sig-nificant differences. The lack of signif-icant deficit in the D group on thephonological pretest requires some ex-planation, and perhaps it can be ex-plained in the context of the theory ofBaddeley and Hitch (1974). The rangeof phonological pretest scores wasgreater for children with dyslexia thanfor typical readers. Some of the highestphonological memory pretest scoresobtained were among participantswith dyslexia. Clearly, not all of thechildren in the D group had deficits inphonological memory span as mea-sured by this test. If the children withdyslexia had phonological deficits,they must have had a means to com-pensate for them, which allowed some

148

\OLLME 35, NUMBER 2, MARCH/APRIL 2002

of them to do as well as the typicalreaders on the phonological pretest.

According to Baddelev's model ofworking memorv (Baddeley, 1986;Gathercole & Baddeley, 1993), a two-part phonological loop supports mem-orv for linguistic material. One part canemplov subvocal articulation (phono-logical rehearsal) to support memoryof verbal material (Baddelev, 1986).The other component is a passive pho-nological storage buffer, specializedfor obligatorv encoding of auditory-verbal input (Salame & Baddeley,1982). Theoreticallv, the passive compo-nent is directlv activated by auditory-verbal input, Nvithout the necessity ofactive rehearsal. If this buffer werefunctional in some readers with dys-lexia, it might allow them to performwell on the pseudoword repetitiontask for short phonological strings(storage time is estimated at 1.5 sec-onds; Baddelev, Lewis, & Vallar, 1984)in spite of their phonological rehearsaldeficits. Children with dyslexia whodid well on the pretest pseudowordrepetition test might be using thispassive mechanism, reflecting recallwithout encoding and rehearsal. Onthe phonological posttest, the higherphonological encoding ability of typi-cal readers apparentlv gave them anadvantage that surpassed anv gainedbv effective echoic memorv in theD group.

If high phonological memory spanperformance in some children withdyslexia is based on articulatory stor-age without rehearsal, these childrenmight be expected to show sharply di-minished performance when the stor-age capacity of the buffer is exceeded.Among children in the D group, thethree highest scorers on the phonolog-ical pretest were able to correctly re-produce most of the elements of thefour-pseudoword lists but less than45% of the phonological elements inthe five-pseudoword lists. Two of thesechildren were not willing to attemptthe six-element lists, and one tried butdid not correctlv repeat anv of thephonological elements. The three high-

Visual 10.0score

9.5

9.0

8.5

8.0

7.5

7.0

6.5

6.0

5.5

5.0

4.5

4.0

primacy middb

Senial posilon

FIGURE 5. Visual memory span pretest and posttest serial position scores for allgroups.

est scoring typical readers, who re-ceived similar scores on the phonolog-ical pretest, tried the six-element lists,correctly repeating 85%, 41 %, and 41 %of the phonological elements on theselists (see Table 8).

Developmental studies have shownthat the ability to read tends to developconcurrently with the ability for pho-nological rehearsal (Walker, Hitch,Doyle, & Porter, 1994). Phonologicalprimacy is taken to represent the roleof phonological rehearsal in support-ing memory for verbal material, whichis essential for fluent reading (Badde-ley, 1986; Baddeley & Hitch, 1974;Hulme & McKenzie, 1992; Spring &Capps, 1974). Individuals have themost time to rehearse the earliest list

items, so they will remember thembest. As young children are learningto read, their primacy performanceshould improve.

It was expected that phonologicalmemory span scores would be posi-tively related to reading scores. Sig-nificant relationships were obtained,especially in children with typicalreading skills, between phonologicalprimacy (as measured by primacy ef-fects on the nonword repetition test)and reading scores. This provides sup-port for the premise that phonologicalrehearsal ability is related to readingdevelopment. The lower scores onphonological pretest primacy obtainedby younger beginning readers in theD-RL group reinforce this conclusion,

reoency

* pree

* poottest

*= dysexeda

- - - = compensated

___ = contiro

149

'15 JORAL OF LEARNING- DI w-S -ABvLi TIEvS-v

TABLE 8Comparison of Phonological Pretest Scores for Highest Scorers Among

Children with Dyslexia and Typical Readers by Number ofElements in List Recalled Correctly

D group participant MC group participant

Score 1 2 3 4 5 6

Total 126 122 90 123 122 111

3-element list 38 36 35 37 38 35

4-element list 50 49 39 47 44 39

5-element list 30 26 28 55 39 26

6-element list 0 0 0 58 28 28

7-element list 0 0 0 11 0 0

Note. D = children with dyslexia; MC = controls matched on chronological age or reading level.

as these Grade 1 readers are at an agewhen rehearsal skills are thought to bedeveloping (Hayes & Schulze, 1977;Hitch, Halliday, Schaafstal, & Heffer-nan, 1991; Hitch, Halliday, Schaafstal,& Schraagen, 1988; Walker et al., 1994).The laborious processing of text ob-served in beginning readers has beenattributed to deficient rehearsal skill.

Phonological primacy scores werecompared to phonological middle po-sition scores to determine whether theprimacy effects (higher score for be-ginning list items) were significant.There were no significant primacy ef-fects for the phonological pretest in theD group, but pretest primacy effectswere significant in all other groups.Phonological posttest primacy effectswere significant in all groups. This sug-gests that phonological rehearsal mayhave been a memory strategy usedonly by typical readers and children inthe C group on the phonologicalpretest, and by all participants on thephonological posttest.

Children with dyslexia may not havebeen able to use phonological rehearsalas an effective memory strategy on thephonological pretest due to encodinglimitations for the novel sound combi-nations. They may have relied onechoic memory instead, at least for theshorter sequences. As the sounds be-came more familiar through the sound-symbol training, encoding them mayhave become easier for the children

with dyslexia. On the phonologicalposttest, children with dyslexia showedreduced primacy effects relative totheir matched peers, but the primacyeffects were significant, suggestingthat the children in this group can usephonological rehearsal to supportmemory of verbal material. This is con-sistent with the results of Gathercoleand Baddeley (1990a). The limitationsin phonological memory of the chil-dren in the dyslexia group may havebeen more related to the encoding ofthe novel sound combinations than torehearsal ability.

Following the sound-symbol train-ing procedure, memory span for thesounds was expected to increase intypical readers but not in children withdyslexia. The sound-symbol trainingdid facilitate performance on thephonological task for typical readers aswell as for children with a history ofdyslexia. All groups had significantlyhigher phonological memory spanscores on the posttest than on thepretest. Training interacted signifi-cantly with the phonological scores, re-sulting in significantly greater gainsfor typical readers than for participantsin the D group and in significant dif-ferences in phonological posttestscores between children in the D groupand their age-matched peers. The facil-itative effect of the training was lesspronounced in children with dyslexiathan in typical readers. This may be be-

cause the children with dyslexia hadmore difficulty discriminating, encod-ing, or retrieving phonological ele-ments from memory than their typi-cally reading peers (Badian, 1998;Siegel & Linder, 1984; Siegel & Ryan,1988), but it may also reflect a reducedbenefit from having a visual symbol toassociate with the sound. The increasein the scores from pretest to posttestmay be due to increased familiaritywith the sounds gained through thetraining (i.e., better phonological rep-resentations; Torgesen et al., 2001) andpossibly also to the facilitative effect ofmultiple coding.

Children with dyslexia were ex-pected to have deficits in memory spanfor newly learned sound-symbol asso-ciations in comparison with typicalreaders. Cross-modal memory scores,as represented by the sound-symboltraining procedure used, were relatedto reading across all reading groups.This supports the findings of Snowling(1980) and others (Spector, 1995; Wal-ton et al., 2001) that reading skilldeficits of children with dyslexia areassociated with extreme difficulties inlearning to apply symbol-sound corre-spondence rules.

Training scores also showed signifi-cant group differences between theD group and both its matched compar-ison groups and between the C groupand its age-matched comparison group.Fifty percent of the children withdyslexia, but only 7% of the typicalreaders, did not master the 10 sound-symbol associations during the sound-symbol training intervention. Mauerand Kamhi (1996) also found childrenwith dyslexia to be slower to learnsound-symbol correspondences thantypical readers in their study of mem-ory for visually and phonologicallysimilar and dissimilar correspondencepairs. This is in contrast to Swanson's(1986) report that there was no reading-related difference between groups innumber of trials to criterion on hissound-symbol training procedures.However, Swanson used meaningfulcommon words as names, and thepresent study used pseudowords. The

JOURNAL OF LEARNING DISABILITIES150

VOLUME 35, NUMBER 2. MARCH/APRIL 2002

pseudowords are novel sound combi-nations, so thev were probably harderfor the children to memorize thanfamiliar combinations. Clearly, thiswas especiallv true for children withdyslexia.

It was expected that if associativememorv deficits in readers with dys-lexia are a product of phonological def-icits, phonological memory span wouldbe positivelv correlated with memoryspan for newly learned sound-symbolcorrespondences in the D group. Train-ing scores were significantly related tophonological serial position scores inthe Grade I readers of the D-RL group.The training scores were significantlycorrelated with iconic memory scoresin typical readers. Training scores werenot significantly associated with any ofthe memory variables in the dyslexiagroup. Cross-modal deficits amongchildren with dvslexia are significantand seem to be somewhat separatefrom visual and phonological skills asrepresented by the tests used in thisstudy. This suggests that deficits insound-symbol association memory inthe children with dyslexia are not asimple product of phonological and vi-sual deficits but may be a separate areaof impairment. Cross-modal deficits inthe D group do not seem to merely re-flect deficits in the phonological as-pects of the task, as has been suggestedby other researchers (McDougall et al.,1994; Vellutino & Scanlon, 1982).

Developmental studies of memoryhave indicated that young childrenand beginning readers rely on visualmemory more than on verbal memoryin image recall (Hitch, Halliday, Dodd,& Littler, 1989; Hitch et al., 1991; Hitch,Woodin, & Baker, 1989; Mann & Liber-man, 1984; Paivio, 1986; Stanovich,1986; Walker et al., 1994). Lack of spon-taneous use of a phonological re-hearsal strategy is assumed to be amajor reason for the reliance on visualmemorv observed in younger children(Hayes & Schulze, 1977; Hitch et al.,1991; Hitch et al., 1988; Walker et al.,1994). This may also be true for chil-dren with dvslexia. Reading skills andrehearsal capabilities seem to develop

concurrently, so that in the elementaryyears children rely on verbal ratherthan on visual memory to support therecognition of letters (Walker et al.,1994) and words (Stanovich, 1986).Stanovich (1986) suggested that asreading becomes more skilled, visual-orthographic strategies again predomi-nate, except for some unfamiliar words.

In this study, cross-modal perfor-mance seems to depend on phonologi-cal skills in very young children.Among the Grade I students partici-pating in the study (the D-RL group),only phonological pretest primacyscores (reflecting the rehearsal of novelsound combinations) were signifi-cantly associated with word and pseu-doword reading ability. When theolder typical readers are included inthe correlation, significant associa-tions of reading and other memorymeasures, especially iconic memory,emerge. Sound-symbol training per-formance seems to become more de-pendent on visual skills as children getolder.

Stanovich (1986) defined skills thatare developmentally limited with re-spect to their effect on reading. That is,some skills affect the acquisition ofreading at certain stages but become soautomatic at later stages that they nolonger are limiting factors. Phonologi-cal encoding may reflect a skill that isusually involved in the learning ofsound-symbol correspondences in adevelopmentally limited way. Most ofthe typical readers tested in this studywere probably able to encode the novelsound combinations sufficiently easilythat learning the pseudowords them-selves was not a limitation for them inlearning the sound-symbol associa-tions. Developmentally persistent pho-nological encoding limitations mayaffect the ability of children with dys-lexia to learn novel sound-symbol as-sociations.

If the tendency to use verbal labels tosupport visual memory develops con-currently with beginning reading skills(Hitch et al., 1988; Hitch et al., 1989;Hitch et al., 1991; Walker et al., 1994)and its development is essential for

cross-modal performance, training andphonological scores should be posi-tively associated in the children in theD-RL group. A significant associationof phonological posttest serial positionscores and training scores was ob-served in these beginning readers.

Another hypothesis examined wasthat in comparison to children withtypical reading skills, children withdyslexia would show lower recall ofitems at the beginning of phonologicalsequences but similar recall of items atthe end of sequences. The phonologicalpretest curves for the D group mostclosely resemble those for the readinglevel control group children. Thephonological pretest results suggestthat phonological memory in thegroup with dyslexia is comparable tothat of younger children reading at thesame level. Based on this finding, onecould conclude that the relationship ofphonological memory to reading inchildren with dyslexia can be charac-terized as a developmental delay (Bad-deley, Ellis, Miles, & Lewis, 1982;Bryant & Impey, 1986).

The phonological pre- and posttestsare both tests of phonological memory.The pretest is a measure of memoryspan for novel stimuli, and the posttestmeasures memory for familiar stimuli.Dyslexia and reading level-matchedgroups scored similarly with novelphonological stimuli, but significantlydifferently with the learned stimuli.The D-RL group gained most follow-ing the training, and the D groupgained least. The effect of sound-symbol training on phonological post-test scores refutes the assumption thatphonological memory in the D groupis similar to that in the D-RL group.Phonological memory in the D groupwas much less facilitated by the train-ing, suggesting that the D group chil-dren did not remember the pseudo-words as well as the D-RL children.

Unlike the D group, D-RL groupchildren did show significant primacyeffects in phonological memory, evenon the pretest, so apparently they wereable to encode and rehearse the novelsound combinations. If children need

151

JOURNAL OF LEARNING DISABILITIES

to develop phonological encoding skillto a certain level before they can gainknowledge of sound-symbol corre-spondences, this must happen beforeGrade 1, as the encoding per se doesnot appear to be a limitation for theD-RL children. Lower phonologicalpretest scores in this group thanamong older typical readers may re-flect a limitation in the ability to re-hearse longer series of the sound com-binations, consistent with the results ofWalker et al. (1994). Although thephonological pretest performances ofthe children with dyslexia and theirreading level-matched peers were sim-ilar, the performances seemed to reflectdifferent types of limitations. The chil-dren in the D group seem to be limitedby encoding ability, as reflected by thelack of pretest primacy.

Perhaps the young children in theD-RL group are particularly adept inmastering novel sound combinations.This seems logical, as Grade 1 childrenare at a stage when their vocabulary islikely to be growing fast. The differ-ence between the children with dys-lexia and their reading level-matchedpeers in their response to the trainingintervention suggests a developmentaldeviance in the children with dys-lexia as opposed to a developmentallag. This supports the findings ofO'Shaughnessy and Swanson (1998)and others.

Phonological posttest serial positionscores did not show recency effects(i.e., a significantly higher score at therecency position relative to the middleposition score), as did the pretest scoresin the typically reading and C groups(see Figure 4). Primacy and middle po-sition scores were significantly higheron the phonological posttest than onthe pretest for typical readers and chil-dren whose dyslexia had been com-pensated, indicating more effective re-hearsal of posttest lists. The increase inprimacy scores and the reduction in re-cency effects following training sug-gests that the children were usingphonological rehearsal to support mem-ory but that by doing so they impaired

their ability to recall the most recentstimuli.

The performance of participantswith dyslexia on the phonologicalposttest was lower than that of com-parison groups, and their mean phono-logical posttest recency was lower thantheir pretest recency. Still, significantrecency effects were present on theposttest in this group. The children inthe D group showed a strength inphonological recency relative to theiroverall performance, suggesting strat-egic differences. Strength in phonolog-ical recency in readers with dyslexiahas also been demonstrated in otherstudies (Farnham-Diggory & Gregg,1975; Spring & Capps, 1974).

Coltheart (1980) described a process-ing bottleneck in memory as follows:

Setting up an iconic memory consists oftemporarily attaching various forms ofphysical information to a permanentlyexisting entry in the internal lexicon. Theattachment is a rapid, automatic processof unlimited capacity; but the attachedinformation decays rapidly .... A lexicalmonitor must operate on the physical in-formation, transforming it into somemore durable form. (p. 223)

Coltheart suggested that the lexicalmonitor has a limited capacity, whichmay vary among individuals. The ef-fort to stabilize the phonological con-tents of the passive storage buffer(hypothesized by Baddeley, 1986) byphonological rehearsal may be a pro-cessing bottleneck for children withlow phonological skills.

Visual Memory

In addition to phonological difficulties,children with dyslexia may also havedeficits in visual memory for pseudo-letter forms (Badcock & Lovegrove,1981; Eden et al., 1995; Lovegrove &Brown, 1978; Lovegrove, Billing, &Slaghuis, 1978; Lovegrove, Heddle,& Slaghuis, 1980). Although many au-thors have claimed that visual codingis intact in children with dyslexia(Hulme, 1988; Katz et al., 1981; Shank-

weiler, Liberman, Mark, Fowler, & Fis-cher, 1979; Torgesen, 1988; Vellutinoet al., 1975; Vellutino, Steger, & Kandel,1972), the children in the D groupshowed a significant deficit relative toage-matched comparison students onthe iconic memory pretest. In fact, rel-ative to both its comparison groups,the D group had consistently lowerscores for visual memory for pseudo-letters. The possibility of a strength iniconic memory storage among childrenwith dyslexia was not supported inthis investigation. The children withdyslexia had deficits in visual iconicmemory relative to the age-matchedgroup as well as to the reading level-matched group when age was used asa covariate.

It was expected that scores on visualiconic memory and visual memoryspan would be positively related toreading scores. Visual memory scoresactually presented a more complex pic-ture than phonological scores in theirrelationship to reading. Iconic memorywas significantly higher in children inthe C group than in children in theD group. When the data for D andC groups were combined, the iconicpretest score was significantly relatedto the WRAT-3 score, r(28) = .46, p = .01,and the WRMT-R score, r(28) = .40, p =.03, although this correlation was notsignificant in either of these groupsseparately or in the groups of typicalreaders. Iconic memory pretest scoreswere significantly correlated withword and pseudoword reading (usingpartial correlation, controlling for age),r = .62, p = .001, in the entire sample.This correlation may reflect the signif-icant relationship between iconicmemory and reading observed in theD and C groups but not among typicalreaders. The iconic memory measure,not obviously like a reading task, ap-parently tapped some cognitive skillimportant for reading, especially inchildren with dyslexia.

The association of iconic memoryscores with both word and pseudo-word reading in children with a his-tory of reading disability may reflect

152

VOLUME 33. NUMBER 2, MARCH APRIL 2�i2 153

either a strategic difference or a limit-ing role of the visual skill tapped inreading ability. The iconic memorv testhad a spatial requirement that was notpresent in the visual memorv span test.This mav be the factor underlving theobserved association of iconic memorywith reading, which seems unique tothe children with a history of dvslexia.Previous research has provided evi-dence of spatial memorv deficits inchildren with reading difficulties(Cornoldi et al., 1999; Enns et al., 1995).

Visual memory for pseudoletters, asmeasured bv the iconic memorv task,clearly plays a greater role in wordreading than in pseudowvord reading,as it is not significantlv related toWRMT-R Word Attack scores in thetotal sample. This is logical, as visualmemorv for whole words plavs an im-portant role in fluent reading but couldnot play much of a role in the decodingof novel sound-svmbol combinations.Stanovich (1986) noted that in tvpicalreaders, whole word reading occurs byan automatized visual route ratherthan bv the more laborious process ofletter-by-letter decoding. The correla-tion of iconic memorv scores withword reading might reflect some as-pect of the role that facilitv of, access to,and maintenance of visual representa-tions plav in the use of the visual routefor word reading.

Visual memory span posttest scoresw ere not significantlv associated witheither reading measure in any group,but some of the serial position scoreswere related to reading scores. Visualmemorv span posttest primacy scoreswere significantlv related to wordreading in the entire population, andthe primacv scores were related toword reading but not to pseudowordreading in tvpical readers. Visual mem-orv span posttest middle positionscores were significantlv associatedwith word and pseudoword reading intypical readers. In children with a his-tory of dvslexia (i.e., the D and C groupscombined), there was a significantassociation of visual memorv spanposttest primacy with pseudouword

reading. Maintenance of pseudoletterimages in the presence of retroactiveinterference is presumably a determi-nant factor in performance on tempo-ral primacy and has a logical relation-ship to skilled reading.

Training Effects

Teaching associations between namesand svmbols was expected to increasememory span scores for the symbols intypical readers but not in children withdyslexia. Sound-symbol training didnot result in an improvement in visualmemorv for the visual symbols. Ra-ther, it seemed to do the opposite, morefor typical readers than for childrenwith dyslexia. Possibly, the reductionin performance from pretest to posttestobserved in typical readers and in chil-dren whose dyslexia had been com-pensated represents their ineffectiveefforts to use verbal coding as a mem-orv aid on a predominantly visual task.Other research has provided examplesof nonproductive efforts to use a pho-nological strategy for a task best per-formed visually (Brandimonte, Hitch,& Bishop, 1992a, 1992b; Schooler &Engstler-Schooler, 1990). Typical read-ers are assumed to be flexible in theirchoice of memory strategies, but ver-bal coding seems to predominate andis, at times, used even when it is not themost effective strategy for a memorytask (Brandimonte, Hitch, & Bishop,1992c; Schooler & Engstler-Schooler,1990). "It seems that, whenever possi-ble, verbal recoding of visual stimuli isused, and this affects subsequent vi-sual image processing" (Brandimonteet al., 1992a, p. 165). Readers withdyslexia are believed to have less flex-ibilitv in their choice of strategy (Swan-sonl, 1986).

On the basis of a series of labeltraining experiments, Swanson (1986)claimed that children with dyslexia areunable to combine verbal and visualcodes additively. The results of thisstudv may support Swanson's claim,but with some important differences.In Swanson's studies, children learned

six real word names for filled geomet-ric shapes. There was no reading-related difference in trials to criterion.The names used by Swanson werewords that the children with dyslexiaknew well. In the present study, thenames were pseudowords, and al-though the shapes were more like let-ters, half of the children with dyslexianever reached the mastery criterion ofbeing able to recite all names in onecomplete presentation of the set.

The sound-symbol training was ex-pected to differentially affect visualmemory span serial position scores inchildren with dyslexia and typicalreaders, producing a greater advan-tage in the children with typical read-ing skills. The training did differen-tially affect children with dyslexia andtypical readers, but not in the directionpostulated. In Swanson's (1986) study,name training reduced recall for chil-dren with dyslexia but increased it fortypical readers. In this study, nametraining reduced visual recency scoresfor typical readers and both recencyand middle position scores for C groupchildren but not for the D group.Although this seems to contradictSwanson's results, it is not necessarilyinconsistent with Swanson's conclu-sions. In Swanson's experiments, thetypical readers mastered the codes andused them successfully to support re-call. In this study, the verbal aspect ofthe sound-symbol task was more diffi-cult. Typical readers may have had dif-ficulty in using the sound-symbol as-sociations to support visual memoryspan, and the effort to do so may haveresulted in a decrement in recall. Thisis similar to what happened to the par-ticipants with dyslexia in Swanson'sexperiment. In this study, name train-ing probably had less effect on recallfor children in the dyslexia group be-cause the children in this group did notappear to be using the names to sup-port recall.

Swanson (1986) concluded that labeltraining increases memory for visualforms in typical readers but seems toreduce visual memory in children with

\0UNIF 3;, NL'NIBER 2, MARCH 'APRIL 2002 153

15 JORA OLERNGDSBITS

dyslexia. Phonetic reading methodolo-gies, such as the Orton-Gillinghammethod (Gillingham & Stillman, 1987),the Spalding method (Spalding &Spalding, 1990), and the Slingerlandmethod (Slingerland, 1971), have beenused to teach reading to children withdyslexia. Phonics training is similar tothe label training used by Swanson(1986), in that children are taught toform a verbal association with a visualimage by simultaneous exposure toboth the verbal and visual forms.These remedial methods are multisen-sory in the sense that

our technique is based upon the close as-sociation of visual, auditory and kines-thetic elements.... Each new phonogramis taught by ... processes, which .. . in-volve the association between visual (V),auditory (A) and kinesthetic (K) recordsto the brain. (Gillingham & Stillman,1956, p. 40)

Spalding and Spalding (1990) statedthat "if a child's aural, or ... visual, re-call of letters is weak and vacillating,then the other three avenues to themind reinforce it and strengthen it"(p. 28). If children with dyslexia do notderive a memory benefit from usingmultiple codes, perhaps multisensoryphonics instruction only confusesthem and interferes with their readingskill development. Specific associativememory deficits among children withdyslexia may cause label training pro-cedures such as phonics instruction toreduce their memory for visual stimuli(e.g., written words). The validity ofphonics training as a remedial ap-proach would then be questionable. Inthis study, children in the dyslexiagroup did show an increase in phono-logical memory span following thesound-symbol training, with no signif-icant reduction in visual memory spanfor pseudoletters. So their decodingskills increased with no significant re-duction in visual memory as tested.

Obviously, the novel sound-symbolrelationships presented in this studywere not overlearned to the point ofmastery, and if they had been, a differ-ent pattern of results might have

emerged (Fischer, 1993). But in thiscase, visual memory (essential to wordreading) was reduced after sound-symbol training, and reduced more fortypical readers than for children withdyslexia. The visual-phonetic associa-tions presented in a multisensory pro-gram will not, by analogy, improve vi-sual recognition of words if the soundsand associations are not mastered toautomaticity. In fact, they may inter-fere with whole word learning. This isconsistent with the conclusions ofMeyer, Wood, Hart, and Felton (1998).Reduced fluency as a result of overre-liance on a phonetic strategy could beless of a problem for children withdyslexia than for typical readers, pos-sibly because readers with dyslexiawill not try to use a phonetic strategyas diligently.

The phonological deficits observedin the children with dyslexia in thisstudy seem best characterized as en-coding deficits. These children mayhave difficulty in mastering sound-symbol correspondences because theyhave difficulty in becoming familiarwith the sounds. The extensive repeti-tion and drill of the sounds providedby a remedial phonics program isprobably helpful in making the stu-dents with dyslexia more aware of thecomponent sounds of speech (Torge-sen et al., 2001). If encoding of soundsand sound combinations is the limitingfactor, drill and repetition would sup-port fluency and automaticity.

Clearly, the use of verbal labels forvisual symbols can reduce visual re-sponse accuracy. This has implicationsfor remediation, because in phonicstraining children learn to use sound-symbol associations to read by decod-ing. This may increase their readingability to something approaching flu-ency, but it may also reduce auto-maticity to the detriment of compre-hension (Stanovich, 1986). It wouldseem important, once children havelearned to read by decoding, to em-phasize sight reading of more commonwords and speed reading in general inorder to avoid a reduction in fluencydue to too much decoding. The need to

develop a methodology to increasewhole word reading in support of re-medial phonics instruction has beenemphasized (Badian, 1998; Felton &Pepper, 1995). The results of this studyreinforce this conclusion.

ABOUT THE AUTHORS

Marjorie Gang, MA, is a teacher of studentswith learning disabilities at South Delta Sec-ondary School in Delta, BC, Canada. She alsotrains teachers and tutors at Douglas College inNew Westminster, BC, to work with studentswith learning problems. Her current profes-sional interests are in developing phonics-basedstrategies for remedial reading instruction.Linda S. Siegel, PhD, is a professor in the De-partment of Educational and Counselling Psy-chology and Special Education at the Univer-sity of British Columbia. She holds the DorothyC. Lam Chair in special education. She has pub-lished numerous articles and books in the area oflearning disabilities. Address: Marjorie Gang,2272 Tompkins Crescent, North Vancouver, BCV7H 2E3.

AUTHORS' NOTES

1. This article is based on a master's thesis byMarjorie Gang, accepted by the Departmentof Educational Psychology and Special Edu-cation at the University of British Columbiain April of 1998. The research was partiallysupported by a grant from the Natural Sci-ences and Engineering Research Council ofCanada to Linda Siegel.

2. The authors wish to thank Rachael Bloom-field for her assistance with the data collec-tion.

REFERENCES

Badcock, D., & Lovegrove, W. (1981). Theeffect of contrast, stimulus duration, andspatial frequency on visible persistence innormal and specifically disabled readers.Journal of Experimental Psychology: HumanPerception and Performance, 7, 495-505.

Baddeley, A. D. (1986). Working memory.New York: Oxford University Press.

Baddeley, A., Ellis, N. C., Miles, T. R., &Lewis, V. J. (1982). Developmental andacquired dyslexia: A comparison. Cogni-tion, 11, 185-199.

Baddeley, A. D., & Hitch, G. (1974). Work-ing memory. In G. Bower (Ed.), Recent ad-vances in learning and motivation (pp. 47-89). London: Academic Press.

JOURNAL OF LEARNING DISABILITIES154

VOLU\E 35, NU\MBER 2, MARCH APRIL 2002

Baddelev, A., Lewis, V., & Vallar, G. (1984).Exploring the articulatory loop. The Quaz r-terly Journal of Experimental Psychologi,36A, 233-252.

Badian, N. A. (1998). A validation of the roleof preschool phonological and ortho-graphic skills in the prediction of reading.Journal of Learning Disabilities, 31, 472-481.

Birch, H. G., & Belmont, L. (1964). Auditory-visual integration in normal and retardedreaders. American Journal or Orthopsychia-try, 34, 852-861.

Brandimonte, M. A., Hitch, G. J., & Bishop,D. V .M. (1992a). Influence of short-termmemory codes on visual image process-ing: Evidence from image transformationtasks. Journal of Experimental Psychology:Learning, Memory. and Cognition. 18, 157-165.

Brandimonte, M., Hitch, G., & Bishop, D.(1992b). Manipulation of visual stimuliimpairs mental image transformations.Memory and Cognition, 20. 449-455.

Brandimonte, NM. A., Hitch, G. J., & Bishop,D. V. NM. (1992c). Verbal recoding of visualstimuli impairs mental image transfor-mations. Memornl andl Cognition, 20, 449-455.

Bryant, P., & Impey, L. (1986). The similari-ties between normal readers and devel-opmental and acquired dvslexics. Cogni-tion, 2, 121-137

Ceci, S. J., Lea, S. E. G., & Ringstrom, M. D.(1980). Coding processes in normal andlearning-disabled children: Evidence formodality-specific pathways to the cogni-tive system. Jouriial of Experimeutal Psiy-

chology: Humnan Learning and Memori, 6,785-797.

Coltheart, M. (1980). Iconic memory andvisible persistence. Perception & Psiy,cho-physics, 27, 183-228.

Corkin, S. (1974). Serial ordering deficitsin inferior readers. NeuropsYchologia, 12,347-354.

Cornoldi, C., Rigoni, F, Tressoldi, P. E., &Vio, C. (1999). Imagery deficits in non-verbal learning disabilities. Journal ofLearning Disabilities, 32, 48-57.

Di Lollo, V., Hanson, D., & Mclntyre, J. S.(1983). Initial stages of v isual informationprocessing in dyslexia. Jourtal of Experi-mental Psychology: Humnan Perception andPerformance, 9. 923-935.

Dixon, P., & Twillev, L. C. (1988). Locationconfusions in v-isual information process-ing. Canadian Journal of Psychology. 42,378-394.

Eden, G. F, Stein, J. F., Wood, H. M., &Wood, F. B. (1995). Temporal and spatialprocessing in reading disabled and nor-mal children. Cortex, 31, 451-468.

Enns, J. T., Bryson, S. E., & Roes, C. (1995).Search for letter identity and location bydisabled readers. Canadian Journal of Ex-perimental Psychology, 49, 357-367.

Enns, J. T., & Rensink, R. (1992). VScopeT'':

Vision testing softvare for the Macintosh.Vancouver: Micropsych Software.

Farmer, M., & Klein, R. (1995). The evidencefor a temporal processing deficit linked todyslexia: A review. Psychonomic Bulletin &Rev-iew, 2, 460-493.

Farnham-Diggory, S., & Gregg, L. W. (1975).Short-term memory function in youngreaders. Joirnal of Experimental Child Psy-cholopgy 19. 279-298.

Felton, R. H., & Pepper, P. P. (1995). Earlyidentification and intervention of phono-logical deficits in kindergarten and earlyelementary children at risk for readingdisability. School Psychology Review. 24,405-414.

Fischer, P. E. (1993). Th1e soun1ds and spellingpatterns of Englishi: Phonics for parents andteachers. Framingham, ME: Oxton House.

Foorman, B. R., Francis, D. J., Fletcher, J. M.,& Lynn, A. (1996). Relation of phonolog-ical and orthographic processing to earlyreading: Comparing two approaches toregression-based, reading-level-match de-signs. Journal of Eduicational Psychologiy 88,639-652.

Foorman, B. R., Francis, D. J., Fletcher, J. M.,Schatschneider, C., & Mehta, P. (1998).The role of instruction in learning to read:Preventing reading failure in at-risk chil-dren. Journal of Educational Psychologi, 90,37-55.

Galaburda, A., & Livingstone, M. (1993).Evidence for a magnocellular defect indevelopmental dyslexia. Annals of theNewe York Academy of Sciences, 682, 70-82.

Gallagher, A. M., Laxon, V., Armstrong, E.,& Frith, U. (1996). Phonological difficul-ties in high functioning dyslexics. Readingand Writing: An Interdisciplinary Journal. 8,499-509.

Gathercole, S. E., & Adams, A.-M. (1993).I'honological working memory in veryyoung children. Developmental Psychol-

oyy, 29, 770-778.

Gathercole, S. E., & Baddeley, A. D. (1989).Evaluation of the role of phonologicalSTM in the development of vocabulary inchildren: A longitudinal study. Journial ofMemory and Language. 28, 200-213.

Gathercole, S. E., & Baddeley, A. D. (1990a).Phonological memory deficits in lan-guage disordered children: Is there acausal connection? Journal of Memory andLanguage, 29, 336-360.

Gathercole, S. E., & Baddeley, A. D. (1990b).The role of phonological memory in vo-cabulary acquisition: A study of youngchildren learning new names. British Jour-nal of Psychology, 81, 439-454.

Gathercole, S. E., & Baddeley, A. D. (1993).Working memory and language. Hove, UK:Erlbaum.

Gathercole, S. E., Willis, C., & Baddeley, A.D. (1991). Differentiating phonologicalmemory and awareness of rhyme: Read-ing and vocabulary development in chil-dren. British Journal of Psychology, 82, 387-406.

Gillingham, A., & Stillman, B. (1987). Reme-dial training for children zvith specific dis-ability in reading, spelling, and penmanship.Cambridge, MA: Educators PublishingService.

Hayes, D. S., & Schulze, S. A. (1977). Verbalencoding in preschoolers' serial reten-tion. Child Development, 48, 1066-1070.

Hitch, G. J., Halliday, M. S., Dodd, A., & Lit-tler, J. E. (1989). Development of rehearsalin short-term memory: Differences be-tween pictorial and spoken stimuli.British Journial of Developmental Psychology,7, 347-362.

Hitch, G. J., Halliday, S., Schaafstal, A. M.,& Heffernan, T. M. (1991). Speech, "innerspeech," and the development of short-term memory: Effects of picture-labellingon recall. Joutrnzal of Child Psychology, 51,

220-234.

Hitch, G. J., Halliday, S., Schaafstal, A. M.,& Schraagen, J. M. C. (1988). Visual mem-ory in young children. Memory and Cog-nition, 16, 120-132.

Hitch, G. J., Woodin, M. E., & Baker, S.(1989). Visual and phonological compo-nents of working memory in children.Memory and Cognition, 17, 175-185.

Huba, M. E., & Vellutino, F. R. (1990). Au-ditory and visual retention in poor andnormal readers when verbal encoding isdisrupted. Learnin1g and Individual Differ-ences, 2(1), 95-112.

Hulme, C. (1988). The implausibility oflow-level visual deficits as a cause of chil-dren's reading difficulties. Cognitive Neu-ropsychology, 5, 369-374.

Hulme, C., & Mackenzie, S. (1992). Workingmemory and severe learning difficulties.Hove, UK: Erlbaum.

155

156 JUNLO ERIGDSBLTE

Hulme, C., & Tordoff, V. (1989). Workingmemory development: The effects ofspeech rate, word length, and acousticsimilarity on serial recall. Journal of Ex-perimental Child Psychology, 47, 72-87.

Katz, R. B., Shankweiler, D., & Liberman,I. Y. (1981). Memory for item order andphonetic recoding in the beginningreader. Journal of Experimental Child Psy-chology, 32, 474-484.

Klein, R., & Farmer, M. (1995). Dyslexia anda temporal processing deficit: A reply tothe commentaries. Psychonomic Bulletin &Review, 2, 515-526.

Lovegrove, W., Billing, G., & Slaghuis, W.(1978). Processing of visual contour ori-entation information in normal and read-ing disabled children. Cortex, 14, 268-278.

Lovegrove, W., & Brown, C. (1978). Devel-opment of information processing in nor-mal and disabled readers. Perceptual andMotor Skills, 46, 1047-1054.

Lovegrove, W. J., Heddle, M., & Slaghuis,W. (1980). Reading disability: Spatial fre-quency specific deficits in visual infor-mation store. Neuropsychologia, 18, 111-115.

Mann, V. A., & Liberman, I. Y. (1984).Phonological awareness and verbal short-term memory. Journal of Learning Disabili-ties, 17, 592-598.

Mauer, D. M., & Kamhi, A. G. (1996). Fac-tors that influence phoneme-graphemecorrespondence learning. Journal of Learn-ing Disabilities, 29, 259-270.

McDougall, S., Hulme, C., Ellis, A., &Monk, A. (1994). Learning to read: Therole of short-term memory and phono-logical skills. Journal of Experimental ChildPsychology, 58, 112-133.

Morrison, E J., Giordani, B., & Nagy, J.(1977). Reading disability: An informa-tion processing analysis. Science, 196,77-79.

Meyer, M. S., Wood, E B., Hart, L. A., &Felton, R. H. (1998). Selective predictivevalue of rapid automatized naming inpoor readers. Journal of Learning Disabili-ties, 31, 106-117.

O'Shaughnessy, T. E., & Swanson, H. L.(1998). Do immediate memory deficits instudents with learning disabilities inreading reflect a developmental lag ordeficit: A selective meta-analysis of theliterature. Learning Disability Quarterly,21, 123-147.

Paivio, A. (1986). Mental representations: Adual coding approach. New York: OxfordUniversity Press.

Schooler, J. W., & Engstler-Schooler, T. Y.(1990). Verbal overshadowing of visualmemories: Some things are better left un-said. Cognitive Psychology, 22, 36-71.

Shankweiler, D., Liberman, I. Y., Mark,L. S., Fowler, C. A., & Fischer, F. W. (1979).The speech code and learning to read.Journal of Experimental Psychology: HumanLearning and Memory, 5, 531-545.

Siegel, L. S. (1986). Phonological deficits inchildren with a reading disability. Cana-dian Journal of Special Education, 2(1), 45-54.

Siegel, L. S. (1993). Phonological processingdeficits as the basis of a reading disabil-ity. Developmental Review, 13, 246-257.

Siegel, L. S., & Faux, D. (1989). Acquisitionof certain grapheme-phoneme corre-spondences in normally achieving anddisabled readers. Reading and Writing: AnInterdisciplinary Journal, 1, 37-52.

Siegel, L. S., & Heaven, R. (1986). Catego-rization of learning disabilities. In S. J.Ceci (Ed.), Handbook of cognitive, social,and neurological aspects of learning disabili-ties (Vol. 1, pp. 229-238). Hillsdale, NJ:Erlbaum.

Siegel, L. S., & Linder, B. A. (1984). Short-term memory processes in children withreading and arithmetic learning disabili-ties. Developmental Psychology, 20, 200-207.

Siegel, L. S., & Ryan, E. B. (1988). Develop-ment of grammatical sensitivity, phono-logical, and short term memory skills innormally achieving and learning dis-abled children. Developmental Psychology,24, 28-37.

Siegel, L. S., & Ryan, E. B. (1989). Subtypesof developmental dyslexia: The influenceof definitional variables. Reading andWriting, 1, 257-287.

Siegel, L., Shore, D., & Geva, E. (1995). Evi-dence for superior orthographic skills indyslexics. Psychological Sciences, 6, 250-254.

Slaghuis, W. L., Lovegrove, W. J., & David-son, J. A. (1993). Visual and language pro-cessing deficits are concurrent in dys-lexia. Cortex, 29, 601-615.

Slingerland, B. (1971). A multi-sensory ap-proach to the language arts for specific lan-guage disability children-A guide for pri-mary teachers. Cambridge, MA: EducatorsPublishing Service.

Slosson, R. (1981). Slosson intelligencetest-Revised. New York: Author.

Snowling, M. J. (1980). The development ofgrapheme-phoneme correspondence innormal and dyslexic readers. Journal of

Experimental Child Psychology, 29, 294-305.

Spalding, R. B., & Spalding, W. T. (1990).The writing road to reading: The Spaldingmethod of phonics for teaching speech, writ-ing, and reading. New York: Quill WilliamMorrow.

Spector, J. E. (1995). Phonemic awarenesstraining: Application of the principles ofdirect instruction. Reading and WritingQuarterly: Overcoming Learning Difficul-ties, 11, 37-51.

Spring, C., & Capps, C. (1974). Encodingspeed, rehearsal, and probed recall ofdyslexic boys. Journal of Educational Psy-chology, 66, 780-786.

Stanovich, K. E. (1986). Matthew effects inreading: Some consequences of individ-ual differences in the acquisition of liter-acy. Reading Research Quarterly, 21, 360-403.

Stanovich, K. E. (1988). Explaining the dif-ferences between the dyslexic and thegarden-variety poor reader: The phono-logical core-variable difference model.Journal of Learning Disabilities, 21, 590-603.

Stanovich, K. E., & Siegel, L. S. (1994). Phe-notypic performance profile of childrenwith reading disabilities: A regression-based test of the phonological-corevariable-difference model. Journal of Edu-cational Psychology, 86, 24-53.

Swanson, H. L. (1978). Verbal encoding ef-fects on the visual short-term memory oflearning disabled and normal readers.Journal of Educational Psychology, 70,539-544.

Swanson, H. L. (1983). A study of non-strategic linguistic coding on visual recallof learning disabled readers. Journal ofLearning Disabilities, 16, 209-216.

Swanson, H. L. (1984). Semantic and visualmemory codes in learning disabled read-ers. Journal of Experimental Child Psychol-ogy, 37, 124-140.

Swanson, H. L. (1986). Multiple codingprocesses in learning disabled and skilledreaders. In S. J. Ceci (Ed.), Handbook ofcognitive, social, and neurological aspects oflearning disabilities (Vol. 1, pp. 203-228).Hillsdale, NJ: Erlbaum.

Torgesen, J. K. (1978-1979). Performance ofreading disabled children on serial mem-ory tasks: A selective review of recent re-search. Reading Research Quarterly, 14,58-87.

Torgesen, J. (1988). Studies of children withlearning disabilities who perform poorly

JOURNAL OF LEARNING DISABILITIES156

\OLLME 35, NLMHER 2, MARCH/AI'KIL 2(XJ2 157

on memorv span tasks. Journal of LearningDisabilities, 21, 605-611.

Torgesen, J. K., Alexander, A. W., Wagner,R. K., Rashotte, C. A., Voeller, K. K. S., &Conway, T. (2001). Intensive remedial in-struction for children with severe readingdisabilities: Immediate and long-termoutcomes from two instructional ap-proaches. Journal of Learning Disabilities,34, 33-58.

Torgesen, J. K., Wagner, R. K., & Rashotte,C. A. (1994). Longitudinal studies ofphonological processing and reading.Journal of Learning Disabilities, 27, 276-286.

Torgesen, J. K., Wagner, R. K., Simmons, K.,& Laughon, P (1990). Identifying phono-logical coding problems in disabled read-ers: Naming, counting, or span mea-sures? Learning Disability Quarterly, 13,236-243.

Torgesen, J., & Murphey, H. (1979). Verbalvs. nonverbal and complex vs. simple re-sponses in the paired associate learningof poor readers. The Journal of General Psy-chology, 101, 219-226.

Vandervelden, M., & Siegel, L. S. (1995).Phonological recoding and phonemeawareness in early literacy: A develop-mental approach. Reading Research Quar-terly, 30, 854-875.

Vellutino, F. R., & Scanlon, D. M. (1982).Verbal processing in poor and normalreaders. In C.J. Brainerd & M. Pressley(Eds.), Verbal processes in children: Progressin cognitive development research (pp. 59-93). New York: Springer Verlag.

Vellutino, F. R., Steger, J. A., Harding, C. J.,& Philips, F. (1975). Verbal and non-verbal paired associates learning in poorand normal readers. Neuropsychologia, 13,75-82.

Vellutino, F R., Steger, J. A., & Kandel, G.(1972). Reading disability: An investiga-tion of the perceptual deficit hypothesis.Cortex, 8, 106-118.

Walker, P., Hitch, G. J., Doyle, A., & Porter,T. (1994). The development of short-termvisual memory in young children. Inter-national Journal of Behavioral Development,17(1), 73-89.

Walton, P D., Walton, L. M., & Felton, K.(2001). Teaching rime analogy or letter re-coding reading strategies to prereaders:Effects on prereading skills and wordreading. Journal of Educational Psychology,93, 160-180.

Wilkinson, G. S. (1993). Wide range achieve-ment test (3rd ed.). Wilmington, DE: Jas-tak.

Woodcock, R. W. (1987). Woodcock readingmastery tests-Revised. Circle Pines, MN:American Guidance Service.

% ULUMh 35, NUMBEK 2, %IAK(-H/A]'RIL 2002 157

COPYRIGHT INFORMATION

TITLE: Sound-symbol learning in children with dyslexiaSOURCE: Journal of Learning Disabilities 35 no2 Mr/Ap 2002

WN: 0206002284005

The magazine publisher is the copyright holder of this article and itis reproduced with permission. Further reproduction of this article inviolation of the copyright is prohibited. To contact the publisher:http://www.proedinc.com/.

Copyright 1982-2002 The H.W. Wilson Company. All rights reserved.