Working memory capacity training for elementary school ... · habilidade de leitura de palavras homófonas, além de mostrarem melhor desempenho na leitura de outros seis tipos de

UNIVERSIDADE FEDERAL DE SANTA CATARINA

CENTRO DE COMUNICAÇÃO E EXPRESSÃO

DEPARTAMENTO DE LÍNGUA E LITERATURA ESTRANGEIRAS

Working memory capacity training for elementary school children: which specific

reading skills show the most significant improvements?

por

Márcia Helena Bragio Flores

Orientadora: Profa. Dra. Mailce Borges Mota

Co-orientador: Dr. Lidiomar José Mascarello

Florianópolis, 28 de junho de 2017.

Working memory capacity training for elementary school children: which specific reading skills show the most significant improvements?

Trabalho de Conclusão de Curso (TCC) para a Disciplina LLE7462 do Departamento de Língua e Literatura Estrangeira do Centro de Comunicação e Expressão da Universidade Federal de Santa Catarina – Curso de Letras – Língua Inglesa e Literaturas, como requisito parcial para obtenção do título de Bacharel em Letras – Língua Inglesa e Literaturas.

Orientadora: Drª Mailce Borges Mota

Co-orientador: Dr. Lidiomar Mascarello

BANCA EXAMINADORA

----------------------------------------------------------

Prof. Dr. Celso Tumolo

---------------------------------------------------------

Dr. Gustavo Lopez Estivalét

Florianópolis

July, 2017

ACKNOWLEDGEMENTS

I would like to thank everyone who helped me in the process of going back to

College after so many years. I am grateful to my husband Eduardo, my daughter Kenile

and her kids - my grandson Mikael and my granddaughter Giovanna - and my son

Daniel. I also thank my mother and my brothers for their support. I also want to thank

all my professors at UFSC. In particular, I would like to thank my advisor Mailce

Borges Mota for her supervision and guidance in the writing of this TCC, as well as Dr.

Lidiomar Mascarello for sharing his data with me.

ABSTRACT

Working memory capacity training for elementary school children: which specific

reading skills show the most significant improvements?


Universidade Federal de Santa Catarina/2017

Profa. Dra. Mailce Borges Mota

Dr. Lidiomar Mascarello

Many children face difficulties during the process of learning to read and that seems to

be particularly true within the Brazilian public schools’ educational context.

Working memory capacity (WMC) has been associated with learning processes such as

reading ability and academic achievement. The present study draws on Mascarello

(2016), who carried out an experiment with second graders from a Brazilian public

school aiming at verifying the impact of WMC training on the learning of reading skills.

The results reported in Mascarello (2016) show that the training exerted positive effects

on children’s WMC and on their reading skills. The purpose of the present study is to

detail which specific reading skills showed the most significant improvements in

Mascarello (2016). The analysis carried out for the purposes of this undergraduate paper

showed that, after WMC training, participants of the experimental group improved their

ability to read homophones by 47.74 %, besides also showing better performance in

reading six other kinds of words.

Key-words: working-memory capacity; reading skills; training; learning difficulties;

Number of words: 7.203

Number of pages: 37

RESUMO

Working memory capacity training for elementary school children: which specific reading skills show the most significant improvements?


Universidade Federal de Santa Catarina/2017

Profa. Dra. Mailce Borges Mota

Dr. Lidiomar Mascarello

Muitas crianças encontram dificuldades durante o processo de alfabetização e isto

parece ser particularmente verdadeiro dentro do contexto educacional das escolas

públicas brasileiras. A capacidade de Memória de Trabalho (CMT) está associada a

processos de aprendizagem como a aprendizagem da leitura (alfabetização),

compreensão leitora e rendimento escolar. O presente estudo baseia-se no trabalho de

Mascarello (2016), que realizou um experimento com alunos da segunda série do ensino

fundamental oriundos de uma escolar pública brasileira com o objetivo de verificar o

impacto do treinamento da CMT na aprendizagem de habilidades de leitura. Os

resultados relatados por Mascarello (2016) mostraram que o treinamento exerceu efeitos

positivos na CMT das crianças, assim como em suas habilidades de leitura. O objetivo

do presente estudo é detalhar quais habilidades de leitura específicas mostraram

progressos mais significativos no estudo de Mascarello (2016). A análise realizada para

os propósitos deste Trabalho de Conclusão de Curso mostrou que, após o treinamento

da CMT, os participantes do grupo experimental melhoraram em 47.74 % sua

habilidade de leitura de palavras homófonas, além de mostrarem melhor desempenho na

leitura de outros seis tipos de palavras.

Palavras-chave: memória de trabalho; habilidades de leitura; treinamento, dificuldades de aprendizagem.

Número de palavras: 7.203

Número de páginas: 37

Table of Contents 1. Introduction………………………………………………………………………1 1.1 Significance of the study…………………………………………………………4 2. Review of Literature……………………………………………………………...4 2.1 Memory Systems and Working Memory………………………………………...5 2.2 Working Memory, learning, Uta Frith’s reading model and TCLPP……………8 2.3 WMC Training in Mascarello’s study and its results…………………………...11 3. Method……………………………………………………………………….…15 4. Results and Discussion……………………………………………………….…16 5. Final Remarks……………………………………………………………….….27 6. References………………………………………………………………….…..29

1

1. Introduction

Working memory capacity (WMC) has been shown to be closely related to

children's learning difficulties in reading (e.g., Smith-Spark and Fisk, 2007; Swanson

and Jerman, 2007; Gathercole and Alloway, 2008). According to Kamhi and Pollock

(2005), phonological processing, which includes phonemic awareness, seems to be the

basis for reading decoding skills. As pointed out by Dehn (2008), the role of phonemic

awareness is to enable learners to identify the relationship between sounds and

letters/words, that is, to establish the connection between phonemes and graphemes.

For De Jong (2006), during the early stages of learning to read, learners depend

heavily on phonological short-term memory – a component of working memory (WM).

The author suggests that, together with verbal-working memory, this memory

subcomponent (phonological short-term memory) of WM underlies the process of

decoding in learning to read.

To date, there seems to be no consensus among researchers about how and why

WMC increases. However, the existence of a relation between deficiencies in phonemic

awareness and phonological processing with reading decoding problems, as argued by

Dehn (2008), might help us understand the impact of different interventions that aim at

training WMC.

Research (e.g., Gathercole and Alloway, 2008) shows that, in the process of

learning to read, children with poor WMC struggle in the decoding stage. Thus,

providing early pedagogical intervention to these children could be a suitable strategy to

help them go through the early stages of learning to read (decoding and word reading).

Gathercole and Alloway (2008) state that assessment of WM abilities of children at

early school years can be a reliable source for identifying students with poor WMC.

2

The authors also argue that, following assessment, intervention might be

implemented in order to reduce the adverse effects of poor WMC on learning. Within

the context of many Brazilian public schools, with teachers receiving low wages, poor

material environment and children coming from low income families, assessing

students’ WMC could be a significant step towards the development of a pedagogical

intervention that can minimize the effects of adverse learning conditions.

With a view to contributing to research on the effects of working memory

training on performance, a recent study carried out by Mascarello (2016), under the

supervision of Dra. Mailce Borges Mota, aimed at investigating the effects of an

intervention whose aim was to provide WMC training. The study was carried out with

public school second graders identified as poor readers with below-average WMC

scores. Students were tested by means of the working memory subtest of the WISC-IV

– Wechsler Intelligence Scale for Children IV (Weschler, 2013).

Results from Mascarello’s study showed that, after a 10-week period of regular

WM training activities, participants of the experimental group presented an increase in

WMC and in reading skills at the decoding and word reading levels, as assessed through

the TCLPP - Reading Skill Test for Words and Nonwords (Seabra and Capovilla, 2010).

These results were compared with those of the control group of the study, which

consisted of 28 children also assessed as presenting low WMC and poor reading skills

and who were not submitted to the working memory training. The comparison showed

that there was a significant difference between the two groups’ posttests performance in

terms of WMC and reading skills, with the experimental group achieving the expected

scores for their age and school year. These results were interpreted by Mascarello

(2016) as evidence that the training had a positive effect on WMC scores and that these

effects improved the decoding and word reading skills of the experimental group.

3

Mascarello (2016) shows that individual differences in WMC should be taken

into account when choosing an extra pedagogical support such as WM-based training

intervention for learners with poor WMC. In this scenario, interventions that aim at

enhancing WMC might be an effective support for these children. Efforts to assess and

minimize the effects of underlying factors – such as poor working memory (that may

prevent a child from achieving the basic reading skills) - might be necessary to help

them go through the process of learning to read.

WMC is argued to be related to reading ability (de Jong, 1998; Gathercole et al.,

2004). Although there is no consensus among researchers on whether WMC can be

improved by training, there is evidence to support this claim. For instance, a study by

Holmes et al. (2009), besides testing children from 6 to 11 years of age for WMC,

implemented a 5 to 7 week training program with the children of the experimental

group, who achieved significant improvements in WMC after the training.

Readdressing the results of Mascarello (2016), the present study aimed at

examining which specific reading skills reported the most significant gains among

children of the experimental group. After assessment of working memory capacity and

reading skills, Mascarello divided participants in five groups: experimental (poor

readers with low WMC who received WMC training), control (poor readers with low

WMC who did not receive WMC training), regular (average readers with average

WMC), LWMCregreaders (average readers with low WMC) and AWMCpooreaders

(poor readers with average WMC).

In order to fulfill the objective of the present study, both pre- and posttests

results of each reading test of four groups of Mascarello’s study are compared. The

experimental group’s results are compared first with the control group and then with the

4

LWCregreaders group’s. The results of the regular group are also presented to depict the

expected average performance for the age.

1.1 Significance of the study

Many children face difficulties during the process of learning to read and that

seems to be particularly true within the Brazilian public schools’ educational context.

Nevertheless, some of those children might be struggling with learning to read due to

low WMC. Here I present further results of an intervention that reported positive effects

of WMC training on children’s reading performance. The present study might help to

specify which aspects of reading skills were tackled by the working memory

intervention. This analysis also compares the experimental group’s results and the

control group’s results in each of the TCLPP seven subtests and verifies the percentage

difference between them. Through this comparison, a better understanding of the factors

that underly participants’ improvement in reading performance might be achieved.

Additionally, results from another group of children (LWMCregreaders), who were

assessed as good readers with low working memory scores, are taken into account as

they seem to indicate an important relation between WMC training and the development

of reading skills.

2. Review of literature

In this section, first the human memory systems are briefly described. Second,

the working memory system is presented. Despite the existence of various theoretical

models of working memory and the fact that this study is mainly based on the multiple-

5

component working memory model proposed by Baddeley and Hitch (1974; Baddeley,

1986, 2000, 2007), some aspects of the model proposed by Cowan (1995) are also

mentioned. Third, the role of working memory in complex activities such as speaking

and reading are posed, along with the role of working memory in learning to read. Still

in the same subsection, Uta Frith’s reading model and a view of TCLPP are also

presented. Fourth, Mascarello’s WMC training and its results are described. Finally, a

qualitative analysis of the reading skills in which the experimental group presented the

most significant improvements is presented, followed by a discussion of the results of

this analysis.

2.1 Memory Systems and Working Memory

The concept of human memory is complex, but some of its features are widely

accepted among researchers. Mota (2015) states that memory systems are classified

according to different features, such as retention of information, the duration of memory

in time, quality of information, and episodic and semantic distinctions. The author states

that, despite the different criteria to classify different types of memory, the human

memory is perceived, in all perspectives, as an association of systems rather than a

single system. These systems enable us to register information, maintain it for a certain

amount of time, and recall it when necessary (Squire and Kandel, 2009). Working

memory is part of this alliance of systems.

There is no consensus about the definition of working memory among

researchers, though the relationship between working memory performance and reading

skills has been supported by many studies over the past 30 years (Dehn, 2008).

Moreover, lower levels of reading, such as decoding, seem to be highly related to

6

phonological short-term memory and verbal-working memory, as stated in Swanson et

al. (2006).

For the purpose of this study, the multiple-component model of working

memory proposed by Baddeley and Hitch (1974) – in its reviewed version (Baddeley,

2000) - is taken as the main theoretical framework. The first version of the model

consists of a phonological loop, a visuospatial sketchpad and the central executive as its

main components. In its revised version, the model includes the episodic buffer.

Working memory is a complex construct that is well described in the original

words of its main researcher, Allan Baddeley. He describes working memory as “a

system for the temporary holding and manipulation of information during the

performance of a range of cognitive tasks such as comprehension, learning and

reasoning” (Baddeley, 1986, p.34). Baddeley states that working memory can be

depicted as a multiple-component model composed of the central executive, the

phonological loop, the visuospatial sketchpad and the episodic buffer. The phonological

loop and the visuospatial sketchpad are directly connected with the central executive as

secondary systems, but not between themselves. As a result of years of further research,

in 2000 Baddeley included the episodic buffer as another component of the working

memory system. The episodic buffer is the system responsible for holding information

in a multidimensional code and providing a link between the components of WM and

long term memory and to perception itself.

The central executive is responsible for manipulating information that reaches

the individual either through the phonological loop or the visuospatial sketchpad. An

example of this process happening through the phonological loop, in combination with

the central executive, is when a teacher asks a child to follow some instructions for a

class task. The child has then to keep the steps in mind while trying to do the actual

task. For example: the teacher may ask students to listen to three words and then

7

rehearse them in alphabetic order. To accomplish the task, the child must retain the

three words in mind while manipulating them to find the right order. This manipulation

of information happens through the central executive, whereas input is made available

through the phonological loop.

In sum, Baddeley sees working memory “as a complex interactive system that is

able to provide an interface between cognition and action, an interface that is capable of

handling information in a range of modalities and stages of processing” (Baddeley,

2012; p.18). Figure 1 shows Baddeley´s working memory latest model.

Figure 1: Working Memory (Baddeley, 2011, p.16).

WMC has been associated with learning processes such as reading and academic

achievement (Gathercole et al., 2004). Therefore, some aspects of this association are

presented in the next section, together with a description of Uta Frith’s reading model

and the TCLPP reading test (Seabra and Capovilla, 2010).

8

2.2 Working Memory, learning, Uta Frith’s reading model and TCLPP

As the working memory system is limited in capacity and plays a key role in

many cognitive processes, WMC-based training has been shown to have a significant

impact on reading skills. Minear and Shah (2006) state that even a slight increase in

WMC can have a significant positive effect on children’s academic performances. My

own view is that WMC is indeed linked to the process of learning to read. It seems that

working memory is highly demanded within the phonological stage – as described by

Uta Frith (1985). Furthermore, a disruption in WM components that control attention,

store information and handle different modalities of data coming from perception in

different modes – among other functions- may jeopardize the learning opportunities

faced by children in their daily lives and more specifically during their classroom

activities.

Learning to read is one of the biggest challenges that children face during their

educational trajectory. This long-lasting process has been studied by many researchers

and some reading models have been proposed. Models may foster our comprehension of

such a complex process because they are an attempt to build a theoretical structure to

systematize many different concepts into a broader picture. Uta Frith’s model of reading

acquisition, from 1985, proposes that the learning of reading takes place in three main

stages: the logographic, the alphabetic, and the orthographic. Each of them involves

different reading strategies or paths and relates to the ways children perceive letters and

words throughout the learning process.

Thus, according to Frith (1985), there are three stages involved in the process of

learning to read. The first is the logographic stage, when children “take pictures” of the

9

words or letters as if they were images and, depending on the frequency these images

are seen, children are able to identify them. Visual memory plays an important role in

this phase. As the reading process develops, children enter the second stage, the

alphabetic stage, and learn to decode graphemes into phonemes, which is an attempt to

match letters with sounds. During this stage the phonological route seems to play a key

role in the ongoing process of learning to read. The final stage is the orthographic stage,

which enables children to accurately recognize words. The orthographic stage is reached

when readers have no more need of sounding out words in order to realize their

meaning. At this stage readers can identify a large number of words automatic and

instantly. An internal lexicon built up along earlier stages must be fully accessed now.

Another theoretical approach to reading is the one proposed by Gagné,

Yekovich, and Yekovich (1993), which focuses on skilled reading comprehension. The

authors postulate that the reading process can be divided into 4 levels of

comprehension: decoding, literal comprehension, inferential comprehension, and

comprehension monitoring. The decoding and literal comprehension stages are the

lowest levels of comprehension. The decoding stage is about matching what is being

read to meaning and then accessing the phonological form of what was read (recoding).

Literal comprehension consists of two processes – lexical access and parsing – aims at

conceptual understanding. The inferential comprehension stage involves automatized

decoding skills, conceptual understanding and strategies. This concerns a higher level of

text comprehension and might be employed by fully proficient readers, which is not the

case of children in the early years of elementary school, though these learners may

begin to use different reading strategies. Finally, comprehension monitoing allows

readers to check whether their comprehension is being achieved. This model is

important as it helps to unveil the cognitive process of learning to read.

10

In Mascarello’s study, a reading skill test was applied to participants in order to

determine the stage at which the participants were in the course of learning to read. The

reading test applied was the TCLPP - Teste de Competência de Leitura de Palavras e

Pseudopalavras (Seabra and Capovilla, 2010). The TCLPP reading test consists of 70

items (words and nonwords) plus a set of 8 items for training before taking the actual

test. The items are composed of one image and one word or nonword each, classified in

seven types: correct regular, correct irregular, semantic neighbors, visual neighbors,

phonological neighbors, homophone nonwords (in this study called homophones) and

nonwords.

According to Seabra and Capovilla (2010), the different types of words demand

different reading strategies, allowing for the evaluation of the reading stage of the

participan, in line with Frith’s (1985) . Table 1 shows the association between the

reading model by Uta Frith and the reading skills demanded in the TCLPP

Table 1: Relations between reading skills and TLCPP within Frith’s model. CR for Correct Regular, CI for Correct Irregular, SN for Semantic Neighbours, VN for Visual Neighbours, PN for Phonological Neighbours, N for Nonwords, and HM for Homophones (can be read only through lexical strategy). Reading

stages

Strategy Brain

Hemisphere

Reading Features TCLPP words

and nonwords

Logographic Logographic Right No graphemic composition

perceived

CR, SN, N, CI

Alphabetic Phonological Left Decode word/nonwords IF

regular

CR, SN, N, VN,

PN

Orthographic Lexical Left Decode words/

Nonwords even if nonregular

grapheme/phoneme relation

CR, CI, SN, VN,

PN,

N, HM

From this table, it is possible to see that different reading strategies are

connected to different stages of the process of learning to read, according to Frith´s

11

model. Thus, seven kinds of words – and nonwords – presented in the TCLPP are

distributed in 70 items in order to assess children’s reading skills so that, from the

number of correct answers given by participants – and the kind of words they are able to

identify correctly - children can be assessed on a scale of reading skills. Table 2 presents

TCLPP’s reference classification values.

Table 2: TCLPP’s reference classification values can be seen in the table below. Source: TCLPP (Capovilla & Seabra, 2010).

TCLPP rating scale Below 70 Very Low

From 70 to 84 Low From 85 to 114 Average From 115 to 129 High (H) 130 and above Very High (VH)

TCLPP reading test was applied to Mascarello’s study participants before and

after WMC training. Pre- and post-tests results were the basis to determine children’s

reading stage. Students were also tested by means of the working memory subtest of the

WISC IV – Wechsler Intelligence Scale for Children (Weschler, 2013) and divided into

5 groups according to their scores. Mascarello’s study and the results of the WMC

training are described in the next subsection.

2.3. The results of WMC Training in Mascarello’s study

Data on the assessment of children’s WMC within the Brazilian educational

context is still scarce. To the best of my knowledge, besides the study by Mascarello

(2016), WMC training through games has not been implemented for second-graders in

Brazilian schools. Outside Brazil, an experimental study by Loosli et al. (2011) showed

that working memory training led to significant gains in participant’s performance in

the reading of single words.The authors report that children engaged in a single-based

task and transfer effects were achieved in reading skills.

12

Differently from the abovementioned study, in Mascarello (2016) children

engaged in a multi-game environment training. A description of this original study

follows.

Aiming at investigating the role of working memory in the development of 7-9

years old second graders’ reading skills, in his doctoral dissertation Mascarello (2016)

focused on assessing children of this age group, students at a Brazilian public school.

Participants were assessed for their WMC and reading skills in order to be provided

WMC training. The author investigated the effects of such training on participants’

WMC as well as on their performance in reading.

The training was carried out with 156 second-graders divided into five

subgroups according to the results of WMC and reading skills tests. For assessing

WMC, three main subtests and one complementary subtest of the WISC IV (Weschler,

2013) were used: 1. digits, consisting of presenting numbers in direct order; 2. digits

backwards, consisting of presenting numbers backwards, 3. numbers and letter

sequences, consisting of presenting numbers in ascending order and letters in alphabetic

order and, 4. consisting of arithmetic, consisting of presenting the first five tasks with

pictures/images and the rest as arithmetic problems (orally). According to Mascarello,

the choice for WISC IV was based on the test’s long experimental path, constant

updating, validation, and wide use among psychologists.

For the assessment of reading skills at the decoding and word reading levels, the

TCLPP was the instrument chosen for the study (Seabra and Capovilla, 2010). The most

important TCLPP’s features were presented in the previous section and some more

details are examined later in the discussion section.

13

After pretests the participants were divided into five subgroups according to their

scores on both the reading and the WMC tests (TCLPP and WISC IV, respectively):

experimental (low WMC and poor readers submitted to WMC training), control (low

WMC and poor readers not submitted to WMC training), regular (average WMC and

reading skills), LWMCregreaders (low WMC and average reading skills) and

AWMCpooreaders (average WMC and poor readers).

The experimental group consisted of 31 children with poor working memory and

poor reading skills; the control group consisted of 28 participants with poor working

memory and poor reading skills, the difference between the two groups being

participation in the WMC training; the regular group consisted of 56 participants with

normal scores both for working memory and reading skills and another two groups of

27 and 23 students, respectively, the first encompassing students with poor WMC

(below average levels) but regular reading skills and the latter consisting of poor readers

(below average levels) and regular scores for working memory.

The material in the training program consisted of 17 different games commonly

known by children in the Brazilian context, and were either easily found in stores – at

affordable prices - or could be manufactured by teachers or the children themselves in

kraft paper or similar materials. Some examples of the games are memory cards for

words, memory cards for numbers, cards with the four mathematic operations, and

domino.

The WM training was carried out for 10 weeks, five days per week, one 35-

minute session per day. The whole training consisted of approximately 30 hours. All 17

games were alternately presented to the participants of the experimental group and each

game aimed at specific goals, though sharing a common general objective: processing

14

and keeping information. The specific goals included: training attention skill and

cognitive flexibility, developing linguistic skills, training attention and visuospatial

skills, developing focusing skills, identifying differences and similarities between

images, training phonological skills, elaboration of linguistic strategies, identifying

words that end with the same phoneme, and matching words with the same phoneme.

After the WMC training period, participants of all groups repeated the WMC and

\reading tests and results were compared to those of the pretests in order to investigate

the effects of the training program. Mascarello (2016) pointed out that participants who

had the highest WMC scores also scored the highest in the reading tests. In line with

Gathercole and Alloway (2006), these results support the idea that WMC is related to

reading performance. Additionally, compared to the WMC of the other groups, the

overall WMC of the experimental group had the highest increase in the posttests, as can

be seen in Table 3:

Table 3: Average increase in WM per group. Group WM

pretest scores

WM posttest scores

Average Growth score difference

Experimental 5.83 8.44 2.61

Control 5.43 6.21 0.78

Regular 10.36 10.88 0.52

LWMregularreaders 7.33 8.94 1.61

AWMpoorreaders 8.72 9.11 0.39

The overall results for reading skills are presented in Table 4, where it can be

seen that the experimental group performance is well above the other groups’.

15

Table 4: TCLPP overall results.

Groups Pre-test Post-test % Growth

Regular Group 101.8 107.62 5.72%

Control Group 77.27 79.66 3.09%

Experimental Group 70.19 98.4 40.19%

LWM reg. readers 98.45 100.8 2.38%

Results of the experimental group from the pretests to the posttests, in

percentage average, show a significant growth of 40.19 %, well above the increase

reported for the other groups’ reading performance.

3. Method

Bearing in mind that the present study aimed at determining the reading skills

that reported most significant gains after the WMC training carried out by Mascarello

(2016), the original study was the basis for this work and the data analysed consisted of

participants’ scores on the WM tests and on the TCLPP reading skills test.

Additionally, the literature in the areas of working memory, learning, reading models,

working memory training and the TCLPP was examined in order to reach a better

understanding of the relationship between WMC and the learning of of reading

byelementary school children. Most of this literature is in English and was made

available by my supervisor. To determine which reading skills were the most affected

by the WMC training implemented, I first checked all values related to the scores of all

participants of four groups, keeping the classification used by the original study. The

groups were: experimental (low WMC, poor readers), control (low WMC, poor

readers), regular (average WMC and reading skills) and LWMCregreaders (low WMC

16

and average reading skills). Second, I calculated the percentage growth of the four

groups in each of the seven reading skill tests in order to identify in which specific

reading skill (kinds of words) participants of the experimental group showed the most

significant growth. Finally, I compared the experimental group’s growth with the other

groups’ and discussed the results within the theoretical framework of Frith’s reading

model and the features of each reading stages proposed by this model.

4. Results and Discussion

For the purpose of this study, the results of the TCLPP pre and post reading tests

from the experimental, the control group and the LWM regular readers groups are

analyzed. Additionally, results from the regular group – composed of 56 children

assessed as good readers by the TCLPP test and average WMC by WISC IV – are

occasionaly mentioned for the purposes of comparison as typically developing children.

Despite the improvement reported by all groups in WMC from the pretests to the

posttest scores, a natural marked WMC growth is expected for children in the period

between 5-11 years of age along with a slower growth up to 15 years, when WMC

seems to reach the same level as adults (Gathercole, 2004). The increase in reading

skills noticed in the experimental group surpassed the average growth of the control

group, although both groups consisted of children with low WMC and poor reading

skills.

Therefore, the analysis of the difference between these groups’ performance can

be helpful to comprehend the scope of working memory training and its possible

transfer effect – in this case for reading skills. Furthermore, the specific reading skills

mostly improved amidst groups and the gap between such growths are addressed and

17

may foster comprehension of this issue.

Results from the WMC training described in Mascarello (2016) showed that

children from the experimental group (31 participants) achieved higher percentage

growths in all seven reading subtests encompassed by the TCLPP as compared to

children from the control group (28 participants). According to Capovilla and Seabra

(2010), the average score expected for this age group (7 years old) in the TCLPP

reading test ranges from 85 to 114 points. Before the WMC training, the pre-test overall

score of the experimental group was at 70.19 range whereas the control group was

77.27. The regular group of children with average WMC and assessed as good readers

scored 101.80 points in the pre-test, a score within the average range expected for their

age described in TCLPP’s figures.

Although the experimental group’s results remained below the regular’s –

which were at average levels since the pretests – the experimental group reported

average scores on the posttests in all the seven items (seven different types of

words/nonwords). Meanwhile, children from the control group remained at low levels in

all subtests, except in the ones that they were already at average levels (CR and CI).

Table 4 above shows the TCLPP reading skills test’s classification values in a rating

scale range from below 70 points to 130 and above points. According to the test results,

children are assessed within the average range, below or above it.

Before moving to the analysis of the results, some aspects of the TCLPP test

require evaluation as they are important for a better comprehension of the data available

in Mascarello’s study.

TCLPP’s normatization figures indicate the N (nonwords) subtest to be the

easiest one, as average scores on it were the highest among all subtests, while the

18

homophones are the most difficult ones. This scale of difficulty was built during the

validation process of the TCLPP Reading test, which involved 2,988 Brazilian public

schools children from the first to the fourth grade, being 873 participants from the then

first grade (currently, in the Brazilian educational system this first grade corresponds to

the second grade). Capovilla and Seabra (2010) argue that the level of difficulty of word

reading seems to be determined by the number of possible strategies that can be used.

Hence, the words that can be read through all the three strategies (logographic,

phonological and orthographic) are the easiest; words that require two strategies are of

medium difficulty, and the words that demand the orthographic strategy to be read are

the most difficult (homophones). Table 5 presents the TCLPP difficulty scale.

Table 5: Scale of difficulty of TCLPP reading test, according to normatization. CR for Correct Regular, CI for Correct Irregular, SN for Semantic Neighbours, VN for Visual Neighbours, PN for Phonological Neighbours, HM for Homophones, and N-Nonwords. Easiest Most Difficult (higher scores) (lower scores)

N SN CR VN CI PN HM

In Mascarello’s study, however, the groups scored slightly different in the

pretests. The highest score reported by the control group was 87.66 at the correct

irregular (CI) words subtest, whereas the experimental group scored 87.01 at the

abovementioned subtest. Meanwhile, results from the regular group point to the correct

regular (CR) words subtest as the easiest one with scores at 105.24. However, correct

irregular (CI) words subtest was the only one to report average scores for all the four

groups discussed in the present study. While it is hard to explain the difference between

the validation figures and those of the groups, it is possible to speculate that more

frequent exposure to the words of the CI test by the participants of Mascarello’s study

may have played a role in the results.

19

In order to have a clearer view of the specific reading skills in which

Mascarello’s study participants reported most significant increase (or decrease) in

reading performance, overall pre- and posttests results as well as the results for each test

of the seven kinds of words of the TCLPP are presented and discussed next. Figures

relate to the four groups taken into account for the purpose of the present study.

As already mentioned, the groups were divided in five subgroups according to

the performance of the 156 participants. Here I examined four groups according to their

scores from both the reading and the WMC tests (TCLPP and WISC IV, respectively):

experimental (low WMC and poor readers, submitted to WMC training), control (low

WMC and poor readers, not submitted to WMC training ), regular (average WMC and

reading skills) and LWMCregreaders (low WMC and average reading skills).

Table 4 shows overall TCLPP pre-posttest results from Mascarello’s study

along with the percentage growth of each group. Figures represent average scores

achieved by the participants of each of the four groups examined in the present study in

two moments: pretests (before WMC training) and posttests (applied after the WMC

program was implemented). From this overall results, it is possible to see that the

experimental group’s growth rate from pre- to posttests reaches 40.19 %, well above the

control group (3.09 % of increase) and the LWM regular readers group (2.38 %). Both

the experimental and the control group were at below average range in pretests. While

the experimental group reached the average range expected for the age in posttests

(98.4), the control group remained at below average level (79.66).

Table 6 shows the groups’ reading performance for nonwords. It is important to

notice that participants of the experimental group were below average during pretests

20

and reached the average level expected in the posttests, while the control group

remained below average.

Table 6: Nonwords TCLPP pre- and posttests results and the percentage comparison for four groups.


Regular Group 104.73 105.89 1.11%

Control Group 69.42 81.38 17.23%


LWM reg. readers 103.21 104.81 1.15%

According to Table 6, in the nonwords tests the experimental group increased

performance on the TCLPP by 32.11 % from pretests to posttests, whereas the control

group increased by 17.23 %. The LWM regular readers reported increased by 1.15 %

from pre- to posttests. Once again the experimental group performance changed from

below average to average range. The control group moved from very low (in red) to

below average range (in yellow), according to TCLPP’s rating scale (see Table 5).

Table 7 shows the TCLPP results on the SN subtest. Figures of the experimental

group show that participants reached average level for reading semantic neighbour

words but the LWMregreaders group did not report any growth in this skill and even

showed a slight decrease in score.

21

Table 7: Semantic Neighbours (SN) TCLPP pre- and post-tests results and the percentage comparison for four groups.


Regular Group 100.91 102.96 2.03%

Control Group 72.53 79.34 9.39%


LWM reg. readers 99.92 98.51 -1.41%

In the semantic neighbour (SN) test the experimental group showed a significant

growth of 34.33 % from the pre-test to the post-test, compared to 9.39 % growth of the

control group, 2.03 % of the regular group and a 1.41 % decrease of the LWM regular

readers group.

Additionally, children from the experimental group reached the average range

(98.6 points) expected for the age, while the control group’s results remained at the

below average range (79.34 points), besides reporting a higher score than the LWM

regular readers group in the post-test. Such progress indicates that in this specific

reading skill the experimental group surpassed regular readers.

These results might be evidence that the WMC training implemented had a

positive transfer effect to reading skills, in line with Loosli et al. (2011), who showed

that WM training leads to significant gains in untrained skills such as reading of single

words.

Table 8 presents the TCLPP CR subtest results.

22

Table 8: Correct Regular (CR) TCLPP pre- and posttests results and the percentage comparison for four groups.


Regular Group 105.24 108.16 2.77%

Control Group 85.66 91.15 6.41%


LWM reg. readers 105.36 106.08 0.68%

As can be seen in Table 8, participants of the experimental group were the only

one below average level in pretests. They reached the average level after the WMC

training, as can be seen in their posttest results. This group also showed the most

significant growth in this subtest, as compared to the other groups.

Once again, in the Correct Regular Words (CR) posttest presented in Table 8,

the participants of the experimental group who participated in the WMC training

reported the most significant increase (26.19 %), as compared to the other three groups.

Additionally, these students jumped from below average scores to the average score

level expected for their age. It can be seen that the LWM regular readers group showed

the smallest growth (0.68 %). Such small increase in the CR posttest – as compared to

the other three groups - may suggest that these children might be struggling to progress

in the early stages of learning to read.

Table 9 shows the results of the TCLPP VN subtest. It can be seen that the

control group remained at below average level in the posttest whereas the experimental

group reached average performance. LWMregreaders participants reported no growth in

the same period.

23

Table 9: Visual Neighbours Words (VN) TCLPP pre- and posttests results and the percentage comparison for four groups.


Regular Group 103.43 105.64 2.14%

Control Group 77.57 81.61 5.21%


LWM reg. readers 102.44 102.44 0%

The comparison between the results of the pre-test and post-tests seen in Table 9

once again indicates a significant rise in the experimental group performance in reading

visual neighbour words (23.38 % rise). VN words can be read either through the

phonological or the lexical route, according to Capovilla and Seabra (2010). Therefore,

participants of the WMC training reached the average range scores, whereas the control

group’s remained below average, as assessed by the TCLPP. Control group participants

might still be struggling in the phonological stage whereas children engaged in WMC

training seem to have progressed in their learning process.

Table 10 shows the Correct Irregular (CI) TCLPP pre- and posttests results and

the percentage growth from pretest to posttest for the four groups.

Table 10: Correct Irregular (CI) TCLPP pre- and post-tests results and the percentage comparison for four groups.


Regular Group 102.18 107.96 5.66%

Control Group 87.66 92.65 5.69%


LWM reg. readers 101.49 101.77 0.27%

24

Correct Irregular words can be read using both logographic and lexical

strategies, according to Capovilla and Seabra (2010). Results seen in Table 13

demonstrate that all four groups reached average scores in pre-tests. However, the

experimental group score in the post-test demonstrates that children engaged in WMC

training have surpassed not only the control group’s performance but also the LWM

regular readers group. Additionally, the experimental group reported the most

significant increase in posttest, with 19.51 % increase in this specific skill of reading

correct irregular words.

Table 11 shows results of the phonological neighbor words TCLPP subtest.

Once again the performance of the LWMregreaders group is significantly different from

the experimental group.

Table 11: Phonological Neighbor Words (PN) TCLPP pre- and posttests results and the percentage comparison for four groups.


Regular Group 99.24 104.23 5.03%

Control Group 76.71 82.75 7.87%


LWM reg. readers 97.56 94.80 -2.83%

In the phonological neighbors posttest the experimental group showed 29.02 %

growth against a 7.87 % increase by the control group, as shown in Table 11. The most

significant result arose from the LWM regular readers group reporting a 2.83 %

decrease in average scores from the pretest to the posttest in this skill. In the TCLPP

test, Seabra and Capovilla (2010) argue that the failure in rejecting phonological

neighbor words – up to the age of 9 - might indicate that children do not access the

25

mental lexicon and have problems in phonological processing as well. The results

presented in Table 11 might signal not only that the experimental group reported a

significant progress in this skill, but also that children from the control group, who did

not attend working memory training sessions, were able to perform slightly better in the

posttest than in the pretest.

As a natural growth in WMC is expected between the age of 5 and 11 (Alloway

and Gathercole, 2008), all the groups – except for LWM regular readers - reported an

increase in WMC from pre- to posttests (see Table 3).

In addition, only the children from the LWM regular readers group did not

improve this reading skill, even showing a decrease in WMC scores in the posttest.

Their performance was at least 7.86 % below the lowest progress reported in this test

(from the regular group). This decrease might suggest that this group of children could

have been coping with low working memory, up to this stage, by developing their own

strategies to overcome poor working memory, such as asking for help or using long-

term memory aids. These strategies, among others, are described in Gathercole and

Alloway (2008).

Moreover, the participants of the above mentioned group seem to be “stuck” in

the phonological stage of reading whereas the experimental group showed a significant

progress in this specific skill. Additionally, eight out of the ten PN items in TCLPP are

related to swaps involving the trace of sonority in words, such as “hapelha” (p/b),

“faca” (f/v), “relóchio” (x/j) and “cancuru” (k/g). According to Zorzi (2006), the failure

in identifying the difference between these phonemes might be a sign of problems in the

process of learning to read.

26

Table 12 shows results from the TCLPP homophone words subtest. This was the

result in which the experimental group reported the most significant growth in a specific

reading skill.

Table 12: Homophones (HM) TCLPP pre- and posttests results and the percentage comparison for four groups.


Regular Group 93.02 98.77 6.18%

Control Group 73.17 80.31 9.76%


LWM reg. readers 87.5 92.79 6.04%

Table 12 shows the performance of the four groups evaluated in the present

study from pretests to posttests (after the experimental group has participated in a 10-

week WMC training program. This specific reading skill (homophones) was the one in

which the experimental group showed the most significant increase (47.74 %) among all

the seven subtests of the TLCCP reading test. Additionally, children from this group

jumped from the very low level - in red – to average level, according to Capovilla and

Seabra (2010) rating scale (see Table 2), besides showing impvroved performance in the

posttest, as compared to the LWMregreaders group.

In sum, after engaging in working memory training, children from the

experimental group first assessed as poor readers with poor WMC were no longer

performing as poor readers, as they reached average scores in all reading post-tests.

Additionally, children from the control group – also assessed as poor readers with poor

WMC but who did not attend training – remained below the average range expected in

post-tests in the reading of five specific skills/kinds of words: semantic neighbours

27

(SN), nonwords (N), visual neighbours (VN), phonological neighbours (PN) and

homophones (HM). This group remained at average levels in correct regular and correct

irregular.

5. Final remarks

Soares (2005) states that literacy is the process of learning to read and write

within a structured alphabetic system that depicts reality. Literacy acquisition is a

gradual, complex and challenging process that children go through in the first years of

elementary school.

Evaluating in which stage of the literacy process learners are and the

implications of poor working memory for this process is not an easy task. Although

reading models and assessment tools can be useful in this attempt, reading stages cannot

be perceived as if they were happening independently from one another. There is no

conclusive evidence on how the process of learning to read happens and which exact

routes are used by learners towards achieving reading proficiency. Moreover, the

bounds between the logographic, the alphabetic and orthographic stages cannot be

exactly defined as learners may use different reading strategies alternatively.

However, results from this working memory training provided evidence of its

effectiveness showing that participants from the experimental group achieved the

average scores expected for their age in all reading post-tests. Moreover, the

experimental group surpassed the scores of the control group in all post-test reading

skills besides showing higher scores than the LWM regular readers group in four

specific reading skills: semantic neighbours, correct irregular, phonological neighbours

and homophones.

28

Reading comprehension is a lifelong learning process and it is of fundamental

importance that concepts of literacy, reading models as well as stages and cognitive

processes involved in reading become clearer for teachers and other professional of the

area of education. Moreover, the identification of children’s reading stage and deficits in

WMC might be crucial information for teachers to gird students in pedagogical

activities that enhance the process of learning to read. In addition, interdisciplinary

efforts could also play a significant role in raising awareness on the challenges of

literacy education. It is hoped that more studies like Mascarello (2016) as well as that

presented here are carried out so that we can improve our understanding of the efficacy

of pedagogical interventions aimed at optimizing children’s WMC and, therefore,

fostering academic success.

29

References

Baddeley, A. D., & Hitch, G. (1974). In G. A. Bower (Ed.), The psychology of learning

and motivation (Vol. 8, pp.47-89). New York: Academic.

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

Baddeley, A. D. (2000). The episodic buffer: A new component in working memory?

Trends in Cognitive Sciences, 4, 417-423.

Baddelley, A. D. (2012). Working Memory: Theories, Models, and Controversies.

Annual Review of Psychology, 63,1-29.

Buchanan, L., Westbury, C., & Burgess, C. (2001). Characterizing semantic space:

Neighborhood effects in word recognition. Psychonomic Bulletin &

Review, 8(3), 531-544.

Caravolas, M.; Lervag A.; Defior, S.; Malkova, G. S.; & Hulme, C. (2013). Different

Patterns, but Equivalent Predictors, of Growth in Reading in Consistent and

Inconsistent Orthographies. Association of Psychological Science, 24(8),1398-

1407.

Cowan, N. (2005). Working memory capacity. New York: Psychology.

De Jong, P. F. (1998). Working memory deficits of reading disabled children. Journal

of Experimental Child Psychology, 70, 75-96.

De Jong, P. F. (2006). Understanding normal and impaired reading development: A

working memory perspective. In S. J. Pickering (Ed.), Working memory and

education (pp. 33-60). Burlington, MA: Academic Press.

Dehn, M. J. (2008). Working Memory and Academic Learning: Assessment and

Intervention. Canada: John Wiley.

Engel, P. M. J.; Santos, F. H.; & Gathercole, S. E. (2008). Are working memory

measures free of socioeconomic influence? Journal of Speech, Language, and

Hearing Research, 51(6), 1580-1587.

Engel de Abreu P. M. J.; Puglisi, M. L.; Cruz-Santos, A.; Befi-Lopes, D. M.; Martin, R.

30

(2014). Effects of impoverished environmental conditions on working memory

performance. Memory, 22(4), 323-331.

Engel de Abreu P. M. J. ; Nikaedo, C., Abreu, N. ; Tourinho, C. J. ; Miranda, M. C. ;

Bueno, O. F. A. ; & Martin, R. (2014). Working Memory Screening, School

Context, and Socioeconomic Status: An analysis of the Effectiveness of the

Working Memory Rating Scale in Brazil. Journal of Attentional Disorders,

18(4), 346-356.

Frith, U. (1985). Beneath the surface of developmental dyslexia. In K.E. Patterson, J.

Marshall, and M. Coltheart (Eds.), Surface dyslexia (pp..301-330). Hillsdale, NJ:

Lawrence Erlbaum.

Gagne, E. D., Yekovich, C. W., & Yekovich, F. R. (1993). The Cognitive Psychology of

School Learning (2nd ed.). New York: HarperCollins College.

Gathercole, S. E., Pickering, S. J., Ambridge, B., & Wearing, H. (2004). The structure

of working memory from 4 to 15 years of age. Developmental

psychology, 40(2), 177.

Gathercole, S.E.; Lamont, E., & Alloway, T.P. (2006). Working Memory in the

Classroom. In S. J. Pickering (Ed.), Working memory and education (pp. 219-

240). Burlington, MA: Academic Press.

Gathercole, S. E., & Alloway, T. P. (2008). Working Memory and Learning: A

Practical Guide for Teachers. London: SAGE.

Kamhi, A. G., & Pollock, K. E. (2005). Phonological disorders in children: Clinical

decision making in assessment and intervention. Baltimore: Brookes.

Loosli, Sandra V.; Buschkuehl Martin; Perrig, Walter J.; & Jaeggi, Susanne M. (2011).

Working memory training improves reading processes in typically developing

children. Child Neuropsychology, 26 May, 1-17.

Mascarello, L. (2016). Memória de trabalho e desempenho em leitura: Um estudo com

crianças do ensino fundamental. Unpublished doctoral dissertation. Santa

Catarina: Edufsc.

Minear, M. E. R. E. D. I. T. H., & Shah, P. R. I. T. I. (2006). Sources of working

31

memory deficits in children and possibilities for remediation. Working memory

and education, 273-307.

Mota, M. (2015). Sistemas de Memória e Processamento da Linguagem: Um Breve

Panorama. Revista Linguística/Revista do Programa de Pós-Graduação em

Linguística da Universidade Federal do Rio de Janeiro, 11, 205-215.

Seabra, A. G., & Capovilla, F. C. (2010). Teste de Competência de Leitura de Palavras

e Pseudopalavras: (TCLPP), São Paulo: Memnon.

Squire, L. R., Kandel E. (2009). Memory: from mind to molecule. Second Ed.

Greenwood Village, CO: Roberts & Company.

Smith-Spark, J. H., & Fisk J. E. (2007). Working memory functioning in developmental

dyslexia. Memory, 15, 34-56.

Soares, M. (2005). Alfabetização: acesso a um código ou acesso à leitura. ONG Leia

Brasil. Rio de Janeiro. Disponível em: http://www. leiabrasil. org.

br/leiaecomente/biblioteca_derrubada. htm. Acesso em, 13/06/2017.

Swanson, H. L., & Jerman, O. (2007). The influence of working memory on reading

growth in subgroups of children with disabilities. Journal of Experimental Child

Psychology, 96, 249-283.

Zorzi, J. L. (2006). Alterações ortográficas nos transtornos de aprendizagem. In: Maluf,

M.I. (org.). Aprendizagem: tramas do conhecimento, do saber e da

subjetividade. Rio de Janeiro: Vozes, 144-162.

Weschler, D. (2003). Weschler Intelligence Scale for Children-fourth edition – WISC –

IV. Brazilian adaptation by Rueda, F. J. M.; Noronha, A. P. P.; Sisto, F. F.;

Santos, A. A. A.; & Castro N. R. Ed. NCS Pearson, Inc.

Documents

Working memory capacity training for elementary school ... · habilidade de leitura de palavras homófonas, além de mostrarem melhor desempenho na leitura de outros seis tipos de