Purser 2013 (15) Poor Phonemic Discrimination Does Not Underlie Poor Verbal Short-term Memory in Down Syndrome

8/13/2019 Purser 2013 (15) Poor Phonemic Discrimination Does Not Underlie Poor Verbal Short-term Memory in Down Syndro

1/15

Poor phonemic discrimination does not underlie

poor verbal short-term memory in Down

syndrome

Harry R.M. Purser a,, Christopher Jarrold b

a School of Psychology, Criminology, and Sociology, Faculty of Arts and Social Sciences, Kingston University, Kingston KT1 2EE, UKb School of Experimental Psychology, University of Bristol, Clifton, Bristol BS8 1TU, UK

a r t i c l e i n f o

Article history:

Received 31 October 2012

Revised 24 December 2012

Available online 28 February 2013

Keywords:

Down syndrome

Memory

Short-term memory

Verbal

Language

Phonological

Phonemic discrimination

Recognition

a b s t r a c t

Individuals with Down syndrome tend to have a marked impair-

ment of verbal short-term memory. The chief aim of this study

was to investigate whether phonemic discrimination contributes

to this deficit. The secondary aimwas to investigate whether phono-

logical representations are degraded in verbal short-term memory

in people with Down syndrome relative to control participants. To

answer these questions, two tasks were used: a discrimination task,

in which memory load was as low as possible, and a short-term rec-

ognition task that used the same stimulus items. Individuals with

Down syndrome were found to perform significantly better than a

nonverbal-matched typically developing group on the discrimina-

tion task, but they performed significantly more poorly than that

group on the recognition task. The Down syndrome group was out-

performed by an additional vocabulary-matched control group on

the discrimination task but was outperformed to a markedly greater

extent on the recognition task. Taken together, the results strongly

indicate that phonemic discrimination ability is not central to theverbal short-term memory deficit associated with Down syndrome.

2013 Elsevier Inc. All rights reserved.

Introduction

Individuals with Down syndrome (DS) tend to perform poorly on tests of verbal short-term mem-

ory (Mackenzie & Hulme, 1987; Marcell & Armstrong, 1982; Marcell & Weeks, 1988). This deficit is

0022-0965/$ - see front matter 2013 Elsevier Inc. All rights reserved.

http://dx.doi.org/10.1016/j.jecp.2012.12.010

Corresponding author.

E-mail address: [email protected](H.R.M. Purser).

Journal of Experimental Child Psychology 115 (2013) 115

Contents lists available at SciVerse ScienceDirect

Journal of Experimental Child

Psychologyj o u r n a l h o m e p a g e : w w w . e l s e v i e r . c o m / l o c a t e / j ec p
http://dx.doi.org/10.1016/j.jecp.2012.12.010mailto:[email protected]://dx.doi.org/10.1016/j.jecp.2012.12.010http://www.sciencedirect.com/science/journal/00220965http://www.elsevier.com/locate/jecphttp://www.elsevier.com/locate/jecphttp://www.sciencedirect.com/science/journal/00220965http://dx.doi.org/10.1016/j.jecp.2012.12.010mailto:[email protected]://dx.doi.org/10.1016/j.jecp.2012.12.010http://crossmark.dyndns.org/dialog/?doi=10.1016/j.jecp.2012.12.010&domain=pdf


2/15

reflected by the relatively poor performance of people with DS on the digit span task, in which partic-

ipants listen to a series of digits spoken by an experimenter and then attempt to repeat them back in

serial order. Numerous studies have shown that the digit spans of individuals with DS are poorer than

those of matched control participants (Jarrold & Baddeley, 1997; McDade & Adler, 1980; see also Nss,

Lyster, Hulme, & Melby-Lervg, 2011). Individuals with DS have also been found to perform more

poorly on verbal short-term memory tasks than control participants when matched for performance

on nonverbal short-term memory tasks (Brock & Jarrold, 2004; Jarrold & Baddeley, 1997; Jarrold,

Baddeley, & Hewes, 1999; Jarrold, Baddeley, & Phillips, 2002; Purser & Jarrold, 2005), indicating that

this deficit is specific to the verbal modality.

This deficit in verbal short-term, or phonological, memory may give rise to further cognitive defi-

cits. Baddeley and colleagues (Baddeley, Gathercole, & Papagno, 1998; Gathercole & Baddeley, 1990)

have argued that phonological memory might play a causal role in vocabulary acquisition and lan-

guage comprehension. Although direct evidence for this suggestion in the context of Down syndrome

is rare (Laws, 1998; Mosse & Jarrold, 2011), language abilities, and expressive language abilities in

particular, are poorer in individuals with DS than predicted by their general cognitive abilities (e.g.,

Chapman, 1995, 1997; Fowler, 1990; Nss et al., 2011).

However, Hulme and Roodenrys (1995)argued that, in contrast to Baddeley and colleagues sug-gestion, the verbal short-term memory deficit associated with DS might be a consequence of language

difficulties because performance on phonological memory tasks is influenced by general language

abilities. Successful performance on verbal short-term memory tasks is likely to depend on the ability

to encode phonological representations (cf.Brady, 1997) because one cannot correctly output an item

from memory that was not correct at the input to that memory system. Metsala (1999)argued that

typically developing childrens phonological representations emerge as a consequence of vocabulary

development. In this way, the general language delay shown by individuals with DS could result in

relatively poor phonological discrimination skills and, consequently, poor verbal short-term memory

performance.

Phonological awareness does appear to be an area of relative difficulty for individuals with DS.

Recently, Roch and Jarrold (2008) assessed the phonological awareness skills of a DS group, andreading-matched controls, with three paradigms. One was initial sound detection, where participants

attempted to match a target picture to one of three response pictures on the basis of sharing the same

initial sound (e.g., Which starts with the same sound as beetable, bed, or sun?). Another was

phoneme deletion, which was presented in a similar fashion; the task involved deciding which of

three pictures would match the sound of the target picture following a particular deletion (e.g., If

d is removed from deer, which would matchdoor, eye, or ear?). The remaining task was rhyme

detection, in which participants tried to decide which of three response pictures rhymed with a target

picture. Although the DS group performed more poorly than controls on each phonological awareness

task, the DS group demonstrated particular difficulties on the rhyme detection test, in line with other

studies (Cardoso-Martins, Michalick, & Pollo, 2002; Snowling, Hulme, & Mercer, 2002, see alsoNss,

Melby-Lervg, Hulme, & Lyster, 2012).Another aspect of phonological awareness is phonemic discrimination, sometimes referred to as

segmental awareness. Phonemic discrimination abilities of individuals with DS were investigated by

Brock and Jarrold (2004) with an item discrimination task similar to that developed by Bridgeman

and Snowling (1988). In this task, participants were auditorily presented with pairs of words or non-

words and then asked to respond as to whether the two items were the same or different. The task

performance of the DS group was found to be impaired relative to that of the control participants.

However, there was a possible limitation of the discrimination task used; the participants needed

to hold two items in memory and access the phonemic features of both simultaneously to be success-

ful at the task. Given that individuals with DS show impaired verbal short-term memory, relatively

poor performance on the task may simply reflect poor memory ability. Thus, it remains possible that

individuals with DS do not have a particular impairment of phonemic discrimination ability. Further-more, it is possible that any group that scores poorly on a test of phonemic discriminationor indeed

any measure of phonological awareness more generallymight do so because of poor verbal short-

term memory if the test makes demands on that system (see Ramus & Szenkovits, 2008, for a detailed

discussion of how findings of phonological discrimination can depend on noncentral task demands).

2 H.R.M. Purser, C. Jarrold / Journal of Experimental Child Psychology 115 (2013) 115


3/15

This might be one reason why individuals with DS struggle with rhyme detection, for example. There

is evidence that children with specific language impairment have difficulties in identifying and encod-

ing phonemes, especially with consonant rather than vowel discriminations (Leonard, McGregor, & Al-

len, 1992). Because both children with specific language impairment and individuals with DS show

generally delayed language abilities (Laws & Bishop, 2003), it is possible that individuals with DS have

a similar pattern of difficulties.

In addition to possible atypicalities in the discrimination of vowels and consonants, individuals

with DS may have atypical short-term memory for vowel or consonant information given their

verbal short-term memory difficulties. Results from a study of typical development suggest that,

for verbal stimuli, vowels are more useful than consonants as placeholders in serial memory,

providing the basis for correct ordering. Using CV (consonantvowel) syllables, Drewnowski

(1980) constructed vowel-only lists, in which the order of consonants remained constant across

trials but the order of vowels was altered. In addition, consonant-only lists were constructed,

where only the order of consonants was varied across trials. Order recall of vowel-only lists was

markedly better than that for consonant-only lists, whereas item recall was roughly equivalent

for these two types of lists. Importantly, the superior order recall of vowel-only lists over conso-

nant-only lists was apparent even when the vowel-only lists were more phonologically confusable.Consistent with this notion, Service, Maury, and Luotoniemi (2005) found impaired recall of

vowel-redundant lists (where each list item had the same vowels) but found unimpaired recall

of consonant-redundant lists (see Luotoniemi, Service, & Maury, 2007, for a review of short-term

recall for vowels and consonants).

Given the above findings, the current study had two main aims: (a) to investigate systematically

whether phonemic discrimination difficulties contribute to the verbal short-term memory deficit

associated with DS and (b) to investigate whether phonological representations are degraded in

verbal short-term memory in people with DS relative to typically developing control participants.

A discrimination task was administered, in which memory load was as low as possible (similar to

that used by Jarrold, Thorn, & Stephens, 2009), using the same stimulus items as a subsequent

recognition memory task. In this way, differences across tasks could not be attributed to differencesin stimuli.

There was also a phonemic manipulation to the task, so that the names of the pictures differed

either in a single consonant or a vowel (e.g., a pig and a pin, a coat and a kite). In addition, the verbally

presented word was masked with either a high or low amount of noise. The effect of noise masking

was to make stimuli less discriminable without manipulating phonology. In this way, it was possible

to assess whether individuals with DS are poorer at phonemic discrimination than matched control

participants and, if so, whether this problem is strictly one of phonemic discrimination or a more gen-

eral auditory discrimination problem.

In addition to the discrimination task, a recognition memory task was employed, based on the test

of item memory used by Brock and Jarrold (2004). If individuals with DS were found to be particularly

susceptible to the phonemic manipulation (whether a consonant or vowel was substituted), thiswould imply that the quality of phonemic representations within phonological memory are degraded

in people with DS.

It was important to include a control group matched on nonverbal ability, rather than vocabulary

age, because there is evidence that verbal short-term memory is a causal factor in determining

vocabulary acquisition in young children (Baddeley et al., 1998; Gathercole & Baddeley, 1990). Con-

sequently, matching for performance on vocabulary tasks would bias toward finding group differ-

ences in verbal short-term memory performance. Hence, if a group of 5-year-olds had the same

average vocabulary mental age as a 20-year-old comparison group with a developmental disorder,

one would expect the older group to have a smaller verbal short-term memory span than the youn-

ger group (cf. Jarrold, Baddeley, Hewes, Leeke, & Phillips, 2004). As a result, our approachnamely

matching for nonverbal ability while allowing for a vocabulary advantage in DSwould bias againstfinding a verbal short-term memory deficit in the DS group. Nevertheless, a vocabulary age-matched

control group was also included to ensure that any group differences found between individuals

with DS and nonverbal ability matched controls could not be attributed to differences in verbal

knowledge.

H.R.M. Purser, C. Jarrold / Journal of Experimental Child Psychology 115 (2013) 115 3


4/15

Method

Participants

There were three groups: 14 individuals with DS (DS), 19 typically developing (TD) children

matched for nonverbal mental age, and 19 TD children matched for vocabulary age. All members of

the DS group had confirmed trisomy 21 without mosaicism and were recruited via local support

groups and schools. The TD children were recruited from local schools. Selection of children was on

the basis of being judged at average ability for age by their teachers. All participants had good hearing

and demonstrated a clear understanding of what the task required. All participants were fully in-

formed about the aims of the experiment, had parental consent, and also consented on the day of test-

ing. In a preliminary test session, participants were tested on the British Picture Vocabulary Scale II

(BPVS;Dunn, Dunn, Whetton, & Burley, 1997), a measure of receptive vocabulary. Participants were

also assessed on Ravens Coloured Progressive Matrices (RCPM; Raven, Raven, & Court, 1998), a test

of nonverbal mental age. The DS group had a mean chronological age of 19 years 4 months (19;4 years,

range = 13;726;4, SD= 3;8), a mean vocabulary mental age of 7;11 years (range = 4;1010;7,

SD= 1;6), and a mean RCPM score of 18.6 (range = 1230,SD= 5.1). The nonverbal-matched grouphad a mean chronological age of 6;0 years (range = 5;26;10, SD= 0;6), a mean vocabulary age of

5;11 years (range = 4;38;0, SD= 1;0), and a mean RCPM score of 18.4 (range = 1324, SD= 3.4).

The vocabulary-matched group had a mean chronological age of 8;5 years (range = 8;08;11,

SD= 0;3), a mean vocabulary age of 8;0 years (range = 5;59;4,SD= 1;1), and a mean RCPM score

of 27.7 (range = 1234,SD = 6.2).

A series of Bonferroni-corrected ttests showed that the individuals with DS were significantly older

than the nonverbal-matched group, t(31, Welch-corrected) = 13.09, p< .001, had reliably higher

vocabulary mental ages, t(31) = 4.16, p< .001, but were closely matched for nonverbal mental age,

t(31) = 0.11, p= .912 (without Bonferroni adjustment). The DS group was reliably older than the

vocabulary-matched group, t(31, Welch-corrected) = 10.75,p< .001, with poorer average nonverbal

mental age, t(31) = 4.61, p< .001, but the groups were closely matched for vocabulary age,t(31) = 0.28,p = .781 (without Bonferroni adjustment).

Procedure

Discrimination task

A total of 32 monosyllabic words, in 16 phonologically similar pairs, were used in the experiment.

Of these 16 pairs, 8 differed by one consonant (fish/dish, sun/nun,goat/boat, hair/chair,pig/pin, lock/sock,

box/fox, and ring/king) and 8 differed by the vowel (bell/ball, cat/cot, girl/goal, boy/bee, kite/coat, man/

moon, pan/pen, and wall/well). Stimulus details are given in the Appendix. Stimulus words were re-

corded using Sound Edit 16 (all samples 16 bits, 32 kHz). Two sets of words were recorded: one with

an adult male voice and one with an adult female voice.On each trial, a pair of large cartoon pictures was displayed on a computer touch-screen, each

depicting one member of one of the above word pairs (e.g., fish anddish). After 1 s had elapsed, a

word of 500 ms duration was auditorily presented in a male voice. This word corresponded to

one of the displayed pictures (e.g., either fish or dish). The participant was required to touch the pic-

ture that corresponded to the auditorily presented word. To successfully perform the task, the par-

ticipant needed only to hold a single word in mind long enough to make a single response. In

addition, the two response pictures remained visible until a response was made, reducing memory

demands even further.

The experimental task involved two different types of phonemic distinction (consonant and vowel)

and two levels of noise masking (high and low). These factors were crossed to give four conditions:

consonanthigh, consonantlow, vowelhigh, and vowellow.There were 32 trials of each of these inter-mixed within four blocks. The dependent variable was the number of correct responses. The stimuli

were spread evenly across conditions.

Low noise masking was achieved by mixing a normalized recording of pink noise at 20%

intensity (of the speech stimulus). High noise masking was achieved by mixing each word

http://-/?-http://-/?-


5/15

recording with the pink noise at 80% intensity. Pink noise is weighted, such that each band-

width is equal in power and, thus, masks evenly across the broad frequency spectrum of natural

speech.

Prior to testing, participants were first shown each of the experimental pictures individually and

required to name them, with feedback given as to which words they represented in the forthcoming

experimental task. This was then repeated to ensure that participants were familiar with the verbal

labels ascribed to the pictures. Participants were then given 12 practice trials, balanced across the four

conditions, with feedback as to the correct responses.

The discrimination task was always administered prior to the memory task in a separate

session.

Memory task

Presentation was on a laptop, prominently featuring a cartoon gray mole character to the left of the

screen and a cartoon pink mole character to the right. On each trial, a list of one, two, or three words

were presented auditorily, spoken by the gray mole in a male voice. This was the presentation list. In

a female voice, the pink mole would then speak a sequence of the same number of words, which wasthe recognition list. The presentation of each word lasted 500 ms, with an interstimulus interval (ISI)

of 500 ms. A 1-s interval separated the presentation and recognition lists. The lists were constructed

from the word pairs used in the discrimination task.

On half of all trials, the recognition list was the same as the presentation list. Only noise-masking

level was manipulated on these trials, giving the two control conditions: samehigh and samelow. The

task involved two different types of phonemic distinction (consonant and vowel) and two levels of

noise masking (high and low). These factors were crossed to give four experimental conditions: con-

sonanthigh, consonantlow, vowelhigh, andvowellow. Only the presentation list was masked with

noise; no noise was applied to the recognition list in any condition.

The dependent measure wasA0, a nonparametric analog ofd0.A0 is a measure of discriminability like

d0, but it can be calculated when the participant has a hit or false alarm rate of 1 or 0 in addition to

intermediate values. Unliked0,A0 does not require homogeneous variance. Values ofA 0 vary from 0 to

1, with 0.5 indicating chance performance (seeMcNicol, 1972). Essentially,A 0 is a measure of perfor-

mance that takes response biases into account; therefore, it is particularly suited to yes/no recogni-

tion tasks and is calculated as follows:

A0

1

2 H F

1 H F=4H1 F; 1

whereH is the hit rate (proportion of trials on which the participant correctly detected a difference

between lists) and Fis the false alarm rate (proportion of trials on which the participant incorrectly

reported a difference between lists).

There were 16 trials of one-item list length consisting of 2 consonanthigh (e.g., box/fox, highnoise), 2 consonantlow, 2 vowelhigh (e.g., bell/ball, high noise), 2 vowellow, 4 samehigh, and 4

samelow trials. There were 32 trials of two-item list length made up of 4 consonanthigh, 4 conso-

nantlow, 4 vowelhigh, 4 vowellow, 8 samehigh, and 8 samelow trials. There were 24 trials of

three-item list length composed of 3 consonanthigh, 3 consonantlow, 3 vowelhigh, 3 vowel

low, 6 samehigh, and 6 samelow trials. Over trials where a phonemic substitution was made during

the recognition phase, the substitutions were evenly spread across serial position. The stimuli were

also spread evenly across conditions.

The participant was required to judge whether the pink mole had correctly repeated the gray mo-

les words (She got it right) or whether a word had been substituted (She got it wrong). Responses

were verbal and were recorded into the computer by the experimenters key-press. If a participant

changed his or her mind about a response on a particular trial, it was noted and the response filewas amended after testing.

Participants were given practice trials prior to testing, with 4 trials each at list lengths of one, two,

and three items. Feedback was given as to the correct responses throughout these trials. Across these

12 trials, the above conditions were evenly represented.



6/15

Results

Discrimination task

The data were analyzed using a three-way mixed-design analysis of variance (ANOVA), with with-

in-participants factors of noise masking and phonemic distinction and with a between-participants

factor of group. Post hoc comparisons were paired-samplesttests with Bonferroni correction, with al-

pha level set atp = .016. Descriptive statistics for the groups performance under the different condi-

tions of the discrimination task are given in Table 1. There was a significant main effect of group,

F(1, 49) = 39.34, p< .001, gp2 = .616, due to the DS group performing better than the nonverbal-

matched group and the vocabulary-matched group performing better than both other groups (all

ps < .001). The analysis also revealed a significant main effect of noise masking, F(1, 49) = 341.29,

p< .001, gp2 = .874, due to superior performance with low noise masking, but no significant main ef-

fect of phonemic distinction,F(1,49) < 1,p = .54,gp2 = .008.

There was a reliable interaction of noise masking and group, F(2, 49) = 38.69,p< .001, gp2 = .612. All

groups performed better with low noise than with high noise (all Fs > 50, all ps < .001). For high

noise-masking trials, there was a reliable group effect, F(2, 49) = 44.60, p< .001, gp2 = .645; the DSgroup performed better than the nonverbal-matched group (p< .05), and the vocabulary-matched

group performed better than both other groups (both ps < .001). For low noise-masking trials, there

was also a reliable group effect, F(2,49) = 6.22, p< .01, gp2 = .203; the nonverbal-matched group

performed worse than both the vocabulary-matched and DS groups, but there was no reliable

difference between the DS and vocabulary-matched groups.

There was also an interaction of phonemic distinction and group, F(2,49) = 3.89,p< .05,gp2 = .137.

Post hoc analysis of simple effects revealed that the nonverbal-matched group was marginally better

at discriminating consonant changes over vowel changes, F(1, 18) = 4.00,p = .061, gp2 = .182, whereas

the vocabulary-matched group was marginally better at discriminating vowel changes over consonant

changes, F(1,18) = 3.27, p= .087, gp2 = .154. However, the DS group showed no such difference,

F(1,13) < 1. The interaction, then, arose from two marginal effects in opposite directions coupled witha lack of effect in the DS group.

Furthermore, there was a three-way interaction of group, phonemic distinction, and noise mask-

ing, F(2,49) = 3.89, p< .05, gp2 = .137. There was no significant interaction of phonemic distinction

and noise masking for either the DS group, F(1, 13) < 1, or the nonverbal-matched group,

F(1,18) = 1.74, p= .204, gp2 = .088; however, this interaction was reliable for the vocabulary-

matched group, F(1,18) = 9.87, p< .01,gp2 = .354. Restricting analysis to the high noise-masking tri-

als, the vocabulary-matched group was significantly better at discriminating vowel pairs than con-

sonant pairs, F(1,18) = 8.84, p< .01, gp2 = .329, but on low noise-masking trials, the vocabulary-

matched groups performance did not differ between vowels and consonants, F(1,18) < 1. Thus,

the only effect of phonemic distinction evident in the discrimination task was the vocabulary-

Table 1

Mean numbers of correct responses by group, noise masking, and phonological distinction in the discrimination task.

Group Noise masking Phonological distinction

Vowel Consonant

Mean SD Mean SD

DS High 24.1 3.76 24.1 3.52

Low 31.3 1.23 31.4 0.84

Nonverbal-matched High 21.0 2.40 22.5 2.80

Low 29.9 1.52 30.4 1.54

Vocabulary-matched High 29.5 1.78 28.2 1.98

Low 31.0 1.47 31.1 0.81

Note. Maximum possible score = 32. DS: Down syndrome group; Vocabulary-matched: control group matched on the British

Picture Vocabulary Scale; Nonverbal-matched: control group matched on the Ravens Coloured Progressive Matrices.



7/15

matched groups superior performance with vowels over consonants in high noise. There are sev-

eral possible explanations for this result (e.g., superior phonotactic knowledge in the vocabulary-

matched group facilitating more of a vowel advantage in the absence of ceiling effects). However,

this result does not speak directly to the current research questions and, therefore, is not consid-

ered further.

Recognition task

The data were analyzed using a four-way mixed-design ANOVA, with within-participants factors of

noise masking, phonemic distinction, and list length and with a between-participants factor of group.

Descriptive statistics for the groups performance under the different conditions of the recognition

task are given in Table 2. There was a significant main effect of noise masking, F(1, 49) = 62.19,

p< .001, gp2 = .559, due to superior performance with low noise masking and also a significant main

effect of phonemic distinction, F(1, 49) = 8.58, p < .01,gp2 = .149, owing to superior recognition of vo-

wel changes over consonant changes. There was a reliable main effect of list length, F(2, 98) = 44.16,

p< .001, gp2 = .474. Performance was significantly better with a list length of one item than of two

or three items, and it was also better with a list length of two items than of three items. Furthermore,

there was a reliable main effect of group, F(2, 49) = 109.20, gp2 = .817, due to the DS group performing

more poorly than the nonverbal-matched group and the vocabulary-matched group performing better

than both other groups.

There was a reliable interaction of noise masking and group, F(2,49) = 9.18, p< .001, gp2 = .272;

although both the DS group, F(1, 13) = 14.16, p< .01, gp2 = .521, and nonverbal-matched group,

F(1, 18) = 50.72, p< .001, gp2 = .738, performed reliably better on low noise-masking trials than

on high ones, the vocabulary-matched group showed only a trend in the same direction,

F(1, 18) = 3.702,p= .070, gp2 = .171, perhaps reflecting near-ceiling effects for this group on the rec-

ognition task.

There was a significant interaction of phonemic distinction and group, F(2, 49) = 10.27,p< .001,

gp2 = .295 (seeFig. 1). Post hoc analysis of simple effects revealed that the nonverbal-matched groupwas better at recognizing vowel changes over consonant changes, F(1, 18) = 38.87,p< .001, gp

2 = .683,

but neither the DS group, F(1,13) < 1, nor the vocabulary-matched group, F(1,18) = 2.18, p= .157,

gp2 = .108, showed such a difference. Performance on the consonant contrasts was well above chance

for both the DS group, t(13) = 37.04, p< .001, and the vocabulary-matched group, t(18) = 55.28,

p< .001, which was close to ceiling.

Table 2

A0 values of correct responses by group, noise masking, phonological distinction, and list length position in the recognition task.

Group Noise masking Phonological distinction List length

1 2 3

Mean SD Mean SD Mean SD

DS High Vowel .75 .17 .59 .13 .54 .20

Consonant .59 .12 .63 .12 .53 .16

Low Vowel .81 .14 .69 .15 .59 .13

Consonant .83 .11 .74 .14 .58 .17

Nonverbal-matched High Vowel .78 .11 .68 .13 .61 .11

Consonant .63 .15 .64 .14 .60 .12

Low Vowel .85 .09 .81 .06 .79 .09

Consonant .86 .13 .81 .09 .62 .14

Vocabulary-matched High Vowel .95 .12 .96 .03 .85 .16

Consonant .91 .11 .96 .08 .94 .09

Low Vowel .92 .08 .96 .06 .97 .04

Consonant .98 .02 .96 .05 .93 .08

Note. DS: Down syndrome group; Vocabulary-matched: control group matched on the British Picture Vocabulary Scale;

Nonverbal-matched: control group matched on the Ravens Coloured Progressive Matrices.



8/15

There was a reliable interaction of length and group, F(4,98) = 9.16,p< .001, gp2 = .272. The effect of

group was significant at each list length (all ps < .001), but the pattern of group differences changed

with list length; at a list length of one item, the vocabulary-matched group outperformed both the

DS and nonverbal-matched groups, but there was no reliable difference between the DS and

nonverbal-matched groups. At longer list lengths, the DS group scored lower than both the vocabu-

lary-matched and nonverbal-matched groups, and the vocabulary-matched group outperformed the

nonverbal-matched group. Thus, only longer list lengths distinguished the memory performance of

the DS and nonverbal-matched groups, with the vocabulary-matched group performing consistently

better than the other groups.

There was no significant interaction of noise masking and phonemic distinction, F(1,49) = 1.25,

p= .269, gp2 = .025, or of noise masking and length, F(2,98, GreenhouseGeisser) = 2.27, p= .109,

gp2 = .044. However, there was a significant interaction of phonemic distinction and length,

F(2,98) = 3.71, p< .05, gp2 = .070, which was qualified by a three-way interaction of noise masking,

phonemic distinction, and length, F(2, 98) = 25.30, p< .001, gp2 = .341. The interaction of phonemicdistinction and length was reliable on both high noise-masking trials, F(2, 102) = 12.74, p< .001,

gp2 = .200 (see Fig. 2A), and low noise-masking trials, F(2, 102) = 14.94, p< .001, gp

2 = .227 (see

Fig. 2B). Restricting analysis to the high noise-masking trials, post hoc pairwise comparisons re-

vealed that participants were reliably better at recognizing vowel changes over consonant changes

with a list length of one item, but there was no significant difference at a list length of two or three

items. Restricting the analysis to the low noise-making trials, post hoc pairwise comparisons showed

that participants were significantly better at recognizing vowel changes over consonant changes

with a list length of three items, but there was no reliable difference at a list length of one or

two items.

Comparison of performance across tasks

To directly compare performance on the discrimination task with that on the memory task, partic-

ipants mean scores for each task were converted into z-scores. Each participants performance was

Fig. 1. Phonemic distinction effects by group in the memory task, as assessed by A0 . Vocabulary-matched: control group

matched on the British Picture Vocabulary Scale; Nonverbal-matched: control group matched on the Ravens Coloured

Progressive Matrices; DS: Down syndrome group. Vertical lines depict standard errors of the means.



9/15

standardized against the mean score of each TD group in each task. The z-scores of the DS group areplotted inFig. 3separately for standardization on the vocabulary-matched and nonverbal-matched

groups scores. A two-way mixed-design ANOVA was performed on thez-scores for each standardiza-

tion, omitting the nonrelevant TD group in each case, with a within-participants factor of task and

with a between-participants factor of group. The analysis confirmed that the interaction between

A

B

Fig. 2. (A) List length effects by phonemic distinction on high noise-masking trials, as assessed by A0 . Vertical lines depict

standard errors of the means. (B) List length effects by phonemic distinction on low noise-masking trials, as assessed by A 0 .

Vertical lines depict standard errors of the means.



10/15

group and task was highly significant for both the nonverbal-matched group standardization,

F(1, 31) = 20.72, p< .001, gp2 = .401, and the vocabulary-matched group standardization,

F(1, 31) = 31.60,p < .001, gp2 = .505.

Discussion

There were two main aims of the experiment. One was to determine whether individuals with DS

have particular difficulties with phonemic discrimination, which could lead to poor verbal short-term

memory performance. The other was to investigate whether phonological representations are de-

graded in verbal short-term memory in people with DS relative to TD control participants. A discrim-

ination task was employed, in which participants ability to make phonemic and more general

auditory discriminations was assessed. This task had been designed to make minimal demands on

memory. There was also a recognition task that measured participants memory for lists of words

involving single phonological foils.

Individuals with DS were found to perform significantly better than the nonverbal-matched

group on the discrimination task but performed significantly more poorly than the nonverbal-

matched group on the recognition task. Because the same stimulus words were used in both tasks,

affording a close comparison of phonemic discrimination and verbal short-term memory abilities,

the results of the current study strongly indicate that phonemic discrimination ability is not central

to the verbal short-term memory deficit associated with DS. In addition, because relatively good

hearing was necessary to afford good discrimination task performance, the results also strongly

indicate that the verbal short-term memory deficit in DS is not primarily caused by hearing difficul-

ties. These findings on the recognition task are more striking given the fact that the DS group had a

higher average vocabulary age than this control group. Furthermore, higher average vocabulary age

cannot fully account for the DS groups relatively good performance on the discrimination task; anadditional control group, matched on the BPVS, outperformed the DS group on the discrimination

task. However, this vocabulary-matched group outperformed the DS group to a significantly greater

extent on the recognition task (an interaction was found between group and task; see Fig. 3),

Fig. 3. DS groupz-scores by standardization group and task. Vertical lines depict standard errors of the means. RCPM: DS task

scores standardized on the nonverbal-matched group; BPVS: DS task scores standardized on the vocabulary-matched group.

Vertical lines depict standard errors of the means.



11/15

strengthening the finding that phonemic discrimination ability is not central to poor verbal

short-term memory in DS. Reflecting this verbal short-term memory deficit, the DS group demon-

strated poorer performance on the recognition task on longer lists (i.e., two or three items) than

both control groups.

Regardless of participant group, consonants and vowels were differentially affected by task condi-

tions in the recognition task. Participants recognition of consonants was markedly impaired by high

noise masking at a list length of one item compared with their recognition of vowels. However, in low

noise-masking conditions, this advantage for recognition of vowels over consonants was evident only

at a list length of three items. This suggests that recognition varies as an additive function of both

noise masking and list length but that the representations of consonants are more susceptible to deg-

radation. In this way, consonant recognition was markedly impaired by either high noise masking or a

longer list length, whereas both factors were necessary to see a comparable drop in performance for

recognition of vowels.

In a study comparing recall for vowels and consonants, Surprenant and Neath (1996) showed

that even when vowel-contrasting stimuli were manipulated to render them less discriminable

than consonant-contrasting stimuli, serial recall performance of those vowels was nonetheless

superior for TD adults. Similarly, in the recognition task of the current study, the nonverbal-matched group showed markedly superior recognition of vowel changes over consonant changes.

The vocabulary-matched group was very close to ceiling on the recognition task, so there was no

room for an effect of phonemic distinction in that group. However, the DS group also showed

no vowel advantage despite being comfortably between floor and ceiling. Because there was no

reliable evidence of such group differences in the discrimination taskif anything, the DS group

was better at discriminating vowels than the nonverbal-matched groupthis indicates that the

short-term storage of vowels and consonants in individuals with DS does not operate in the same

way as in TD children.

There are at least two interpretations of the group difference in storage of vowels and conso-

nants. First, individuals with DS might have a dysfunction of auditory sensory memory.Crowder

and Morton (1969)proposed a theory of auditory sensory memory (or the precategorical acousticstore), which assumes that auditorily presented items are stored in a relatively uncategorized

code for a period of approximately 2 s in a modality-specific memory system. Lexico-semantic cod-

ing may be accessed within a few hundred milliseconds of presentation (e.g., Marslen-Wilson &

Welsh, 1978), but auditory sensory memory is assumed not to be influenced by such coding. As

items enter auditory sensory memory, they will interfere with other items within it provided that

the new items are acoustically similar to the previous items. Research has shown that auditory

sensory memory tends to be superior for vowels, which are relatively simple acoustically and of

relatively long duration, than for consonants, which are acoustically more complex and of shorter

duration (Cowan, Lichty, & Grove, 1990; Crowder, 1971, 1973; Darwin & Baddeley, 1974; Pisoni,

1973). The DS groups failure to show superior recall of vowels over consonants in the current

study, then, is consistent with an auditory memory dysfunction. In addition, individuals with DSappear to have a verbal short-term memory that is limited in capacity to perhaps even a single

item (Purser & Jarrold, 2010). Therefore, it is possible that this limited capacity reflects, in part,

an impairment of auditory sensory memory that would otherwise support verbal short-term mem-

ory performance.

Second, individuals with DS might have a degradation of the representations in phonological

memory. If memory representations were degraded, one might expect their subsequent identifica-

tion in memory to be more difficult, so that item memory would be poorer. However, this degrada-

tion of items would also render them less discriminable, so that order errors may also result. Of

course, similar behavior would also result from degraded input to phonological memory as a result

of poor phonemic discrimination (although it is possible that degradation at discrimination and deg-

radation within memory itself could interact nonlinearly; e.g., degraded inputs might further de-grade more quickly within memory than undegraded inputs). Phonological similarity effects in

short-term serial recall appear to be due to an increased incidence of order errors rather than item

errors (Gruneberg & Melton, 1972; Watkins, Watkins, & Crowder, 1974; Wickelgren, 1965), and



12/15

evidence from typical development suggests that vowels are more useful than consonants for cor-

rect ordering of items in verbal short-term memory (Drewnowski, 1980; Service et al., 2005). With

this in mind, if individuals with DS have a particular degradation of vowel information in verbal

short-term memory, one might expect to see an increased incidence of order errors rather than item

errors.

Indeed, there is evidence suggesting that individuals with DS have particular difficulty with

order memory in verbal short-term memory tasks. Purser and Jarrold (2005) carried out a mod-

ified verbal probed recall task involving DS and TD participant groups. In one experiment, where

both item memory and order memory were necessary to afford good recall, the DS group per-

formed significantly more poorly than the control group. However, in a second experiment that

required only item memory, the two groups performed at a similar level to each other. Moreover,

Brock and Jarrold (2004) showed that individuals with DS performed more poorly on a test of

short-term verbal order memory than predicted by their performance on a phonological item

memory task.

Whatever the underlying reason for the DS groups lack of an advantage for recognizing vowel

changes over consonant changes, the current study clearly shows that it does not primarily lie in

difficulties of phonemic discrimination that might be related to the hearing loss typically associatedwith the condition. Moreover, due to the strong interaction found between task and group, our results

indicate that the verbal short-term memory deficit associated with DS cannot be primarily attributed

to such difficulties. This confirms findings from previous work (e.g., Brock & Jarrold, 2004; Marcell &

Cohen, 1992) but builds on them in a number of ways. First, in contrast to previous findings, the DS

group of the current study performed better on the discrimination task than the nonverbal-matched

control group, presumably because of the successful removal of memory load in our task. This

suggests that poor performance by people with DS on other tasks, such as nonword repetition (e.g.,

Cairns & Jarrold, 2005; Laws, 1998), is unlikely to owe primarily to poor phonemic discrimination.

Furthermore, the current study demonstrates the need to minimize memory load in any future inves-

tigations of phonemic discrimination, particularly for participants with etiologies associated with poor

verbal short-term memory.It should be noted that the results of the current study do not imply that poor phonemic discrim-

ination plays no role at all in determining verbal short-term memory performance. The fact that the DS

group performed markedly worse on the discrimination task than the vocabulary-matched group

indicates that phonological processing is weak in DS and is delayed relative to lexico-semantic

knowledge. This phonological processing deficit would be expected to constrain verbal short-term

memory performance, particularly when memoranda are less phonologically distinct (e.g., in school

classrooms;McSporran, 1997).

More general, it might benefit researchers to bear in mind the verbal short-term memory

demands made by various tests used to assess phonological awareness (Gathercole, 2006). For

example, in the oddity task, developed byBradley and Bryant (1978), the participant listens to trip-

lets of words and, on oddity trials, is required to select the odd word out. Clearly, however, the childmust hold in mind at least two words to succeed at the task. Similar concerns clearly apply to

various rhyme tasks and also to synthesis tasks, in which participants must hold in mind the various

subcomponents of a word before joining them together to respond with the target word. One

possible way of addressing these concerns would be to take a concomitant measure of verbal

short-term memory to be used as a covariate in subsequent analysis. Where no attempt has been

made to reduce or account for memory demands in a task of phonological awareness, conclusions

should be viewed with caution.

Acknowledgments

This research was funded by a studentship from the Medical Research Council to Harry Purser and

was also supported by Economic and Social Research Council (ESRC) Grant RES-062-33-0005. We

thank Simon Farrell, Jon Brock, and Clive Frankish for their helpful comments and also Ioanna Angus

and Aalia Javaid for additional data collection.



13/15

Appendix. Details, where available, of words used in the experiment

Word Rating

Rated age

of acquisition(months)

Imageability

(scale 17)

Rated

frequency(scale 15)

Ball 12.0 6.40 3.45

Bee 34.2 6.30 2.85

Bell 40.8 6.60 2.50

Boat 30.6 6.30 3.30

Box 32.4 5.60 3.65

Boy 14.4 6.25 4.10

Cat 12.0 6.40 3.40

Chair 28.8 6.45 4.00

Cot

Dish 52.0 5.94 Fish 32.4 6.75 3.05

Fox 48.0 6.55 2.15

Girl

Goat 46.8 6.30 2.00

Hair 18.0 5.75 4.25

King 37.2 6.35 2.05

Kite 49.2 6.65 1.65

Lock

Man

Moon 34.2 6.65 3.00

Nun 74.4 6.20 1.90Pan 44.4 6.70 3.90

Pen 36.0 6.35 4.45

Pig 23.4 6.75 2.50

Pin

Ring 48.0 5.95 3.45

Sock 23.4 6.20 4.05

Sun 16.2 6.70 3.95

Wall

Well 50.4 4.85 2.20

Note.All ratings are from Morrison, Chappell, and Ellis (1997)except those in bold, which are recent norms from the Universityof Bristol (http://www.pc.rhul.ac.uk/staff/c.davis/Articles/Stadthagen-Gonzalez_Davis_in_press.pdf ).

References

Baddeley, A., Gathercole, S., & Papagno, C. (1998). The phonological loop as a language learning device. Psychological Review, 105,158173.

Bradley, L., & Bryant, P. (1978). Difficulties in auditory organization as a possible cause of reading backwardness. Nature, 271,746747.

Brady, S. A. (1997). Ability to encode phonological representations: An underlying difficulty of poor readers. In B. Blachman

(Ed.),Foundations of reading acquisition and dyslexia (pp. 2147). Mahwah, NJ: Lawrence Erlbaum.Bridgeman, E., & Snowling, M. (1988). The perception of phoneme sequence: A comparison of dyspraxic and normal children.

British Journal of Disorders of Communication, 23, 245252.

Brock, J., & Jarrold, C. (2004). Language influences on verbal short-term memory performance in Down syndrome: Item andorder recognition.Journal of Speech, Language, and Hearing Research, 47, 13341346.

Cairns, P., & Jarrold, C. (2005). Exploring the correlates of impaired nonword repetition in Down syndrome. British Journal ofDevelopmental Psychology, 23, 401416.

Cardoso-Martins, C., Michalick, M. F., & Pollo, T. C. (2002). Is sensitivity to rhyme a developmental precursor to sensitivity to

phoneme? Evidence from individuals with Down syndrome. Reading and Writing: An Interdisciplinary Journal, 15 , 439454.

http://language.psy.bris.ac.uk/norms/BristolNorms(in%20press).pdfhttp://language.psy.bris.ac.uk/norms/BristolNorms(in%20press).pdf


14/15

Chapman, R. S. (1997). Language development in children and adolescents with Down syndrome. Mental Retardation andDevelopmental Disabilities Research Reviews, 3, 307312.

Chapman, R. S. (1995). Language development in children and adolescents with Down syndrome. In P. Fletcher & B.

MacWhinney (Eds.),Handbook of child language (pp. 641663). Oxford, UK: Blackwell.Cowan, N., Lichty, W., & Grove, T. R. (1990). Properties of memory for unattended spoken syllables. Journal of Experimental

Psychology: Learning, Memory, and Cognition, 16, 258269.

Crowder, R. G. (1971). The sound of vowels and consonants in immediate memory.Journal of Verbal Learning and Verbal Behavior,10, 587596.

Crowder, R. G. (1973). Representation of speech sounds in precategorical acoustic storage.Journal of Experimental Psychology, 98,1424.

Crowder, R. G., & Morton, J. (1969). Precategorical acoustic storage (PAS). Perception and Psychophysics, 5, 365373.Darwin, C. J., & Baddeley, A. (1974). Acoustic memory and the perception of speech. Cognitive Psychology, 6, 4160.Drewnowski, A. (1980). Memory functions for vowels and consonants: A reinterpretation of acoustic similarity effects.Journal of

Verbal Learning and Verbal Behaviour, 19, 176193.Dunn, L. M., Dunn, L. M., Whetton, C., & Burley, J. (1997). British Picture Vocabulary Scale II. Windsor, UK: NFERNelson.Fowler, A. E. (1990). Language abilities in children with Down syndrome: Evidence for a specific syntactic delay. In D. Cicchetti &

M. Beeghly (Eds.), Children with Down syndrome: A developmental perspective (pp. 302328). Cambridge, UK: CambridgeUniversity Press.

Gathercole, S. E. (2006). Nonword repetition and word learning: The nature of the relationship. Applied Psycholinguistics, 27,513543.

Gathercole, S. E., & Baddeley, A. D. (1990). The role of phonological memory in vocabulary acquisition: A study of young children

learning new words. British Journal of Psychology, 81, 439454.Gruneberg, M. M., & Melton, K. C. (1972). Acoustic confusions and order forgetting. Acta Psychologica, 36, 4853.Hulme, C., & Roodenrys, S. (1995). Practitioner review: Verbal working memory development and its disorders. Journal of Child

Psychology and Psychiatry, 36, 373398.Jarrold, C., & Baddeley, A. D. (1997). Short-term memory for verbal and visuo-spatial information in Downs syndrome.Cognitive

Neuropsychiatry, 2, 101122.Jarrold, C., Baddeley, A. D., & Hewes, A. K. (1999). Genetically dissociated components of working memory: Evidence from

Downs and Williams syndrome. Neuropsychologia, 37, 637651.Jarrold, C., Baddeley, A. D., Hewes, A. K., Leeke, T., & Phillips, C. (2004). What links verbal short-term memory performance and

vocabulary level? Evidence of changing relationships among individuals with learning disability. Journal of Memory andLanguage, 50, 134148.

Jarrold, C., Baddeley, A. D., & Phillips, C. E. (2002). Verbal short-term memory in Down syndrome: A problem of memory,

audition, or speech? Journal of Speech, Language, and Hearing Research, 45, 531544.Jarrold, C., Thorn, A. S., & Stephens, E. (2009). The relationships among verbal short-term memory, phonological awareness, and

new word learning: Evidence from typical development and Down syndrome.Journal of Experimental Child Psychology, 102,196218.

Laws, G. (1998). The use of nonword repetition as a test of phonological memory in children with Down syndrome. Journal ofChild Psychology and Psychiatry, 39, 11191130.

Laws, G., & Bishop, D. V. M. (2003). A comparison of language abilities in adolescents with Down syndrome and children with

specific language impairment. Journal of Speech, Language, and Hearing Research, 46, 13241339.Leonard, L. B., McGregor, K. K., & Allen, G. D. (1992). Grammatical morphology and speech perception in children with specific

language impairment.Journal of Speech and Hearing Research, 335, 10761085.Luotoniemi, E., Service, E., & Maury, S. (2007). Good and bad effects of phonological similarity on word and nonword recall: The

role of beginnings and ends. European Journal of Cognitive Psychology, 19, 80102.Mackenzie, S., & Hulme, C. (1987). Memory span development in Downs syndrome, severely subnormal and normal subjects.

Cognitive Neuropsychology, 4, 303319.Marcell, M. M., & Armstrong, V. (1982). Auditory and visual sequential memory of Down syndrome and nonretarded children.

American Journal of Mental Deficiency, 87, 8695.Marcell, M. M., & Cohen, S. (1992). Hearing abilities of Down syndrome and other mentally handicapped adolescents. Research in

Developmental Disabilities, 15, 533551.Marcell, M. M., & Weeks, S. L. (1988). Short-term memory difficulties and Downs syndrome. Journal of Mental Deficiency

Research, 32, 153162.Marslen-Wilson, W., & Welsh, A. (1978). Processing interactions and lexical access during word recognition in continuous

speech. Cognitive Psychology, 10, 2963.McDade, H. L., & Adler, S. (1980). Down syndrome and short-term memory impairment: A storage or retrieval deficit?American

Journal of Mental Deficiency, 84, 561567.McNicol, D. (1972). A primer of signal detection theory. London: Allen & Unwin.McSporran, E. (1997). Towards better listening and learning in the classroom. Educational Review, 49, 102111.Metsala, J. L. (1999). Young childrens phonological awareness and non-word repetition as a function of vocabulary

development. Journal of Educational Psychology, 91, 319.

Morrison, C. M., Chappell, T. D., & Ellis, A. W. (1997). Age of acquisition norms for a large set of object names and their relation to

adult estimates and other variables. Quarterly Journal of Experimental Psychology A, 50, 528559.Mosse, E. K., & Jarrold, C. (2011). Evidence for preserved novel word learning in Down syndrome suggests multiple routes to

vocabulary acquisition. Journal of Speech, Language, and Hearing Research, 54, 11371152.

Nss, K.-A. B., Lyster, S.-A. H., Hulme, C., & Melby-Lervg, M. (2011). Language and verbal short-term memory skills in childrenwith Down syndrome: A meta-analytic review. Research in Developmental Disabilities, 32, 22252234.

Nss, K.-A. B., Melby-Lervg, M., Hulme, C., & Lyster, S.-A. H. (2012). Reading skills in children with Down syndrome: A meta-

analytic review.Research in Developmental Disabilities, 33, 737747.Pisoni, D. B. (1973). Auditory short-term memory and vowel perception. Memory & Cognition, 3, 78.



15/15

Purser, H. R. M., & Jarrold, C. (2005). Impaired verbal short-term memory in Down syndrome reflects a capacity limitation rather

than atypically rapid forgetting. Journal of Experimental Child Psychology, 91 , 123.Purser, H. R. M., & Jarrold, C. (2010). Short- and long-termmemory contributions to immediate serial recognition: Evidence from

serial position effects. Quarterly Journal of Experimental Psychology, 63, 679693.Ramus, F., & Szenkovits, G. (2008). What phonological deficit? Quarterly Journal of Experimental Psychology, 61 , 129141.Raven, J., Raven, J. C., & Court, J. H. (1998). Coloured progressive matrices. Oxford, UK: Oxford University Press.

Roch, M., & Jarrold, C. (2008). A comparison between word and non-word reading in Down syndrome: The role of phonologicalawareness. Journal of Communication Disorders, 41, 305318.

Service, E., Maury, S., & Luotoniemi, E. (2005). Forgetting and redintegration of consonants and vowels in pseudoword lists.

Memory, 13, 340348.Snowling, M. J., Hulme, C., & Mercer, R. C. (2002). A deficit in rime awareness in children with Down syndrome. Reading and

Writing: An Interdisciplinary Journal, 15, 471495.Surprenant, A. M., & Neath, I. (1996). The relation between discriminability and memory for vowels, consonants, and silent-

center vowels.Memory and Cognition, 24, 356366.Watkins, M. J., Watkins, O. C., & Crowder, R. G. (1974). The modality effect in free and serial recall as a function of phonological

similarity. Journal of Verbal Learning and Verbal Behavior, 13, 430447.Wickelgren, W. A. (1965). Short-term memory for phonemically similar lists. American Journal of Psychology, 78, 567574.


Documents

Purser 2013 (15) Poor Phonemic Discrimination Does Not Underlie Poor Verbal Short-term Memory in Down Syndrome