Vous êtes sur la page 1sur 20

Development of phonological awareness and

reading acquisition
A study in Spanish language

MAR/SOL CARRILLO
Department of Psychology and Education, University of Murcia, Spain

ABSTRACT. The work is aimed at studying the relations between different levels of phono-
logical awareness and early reading ability. Ten different metaphonological tasks as well as a
reading (syllables and word decoding) test were administered to kindergarteners and first graders.
The correlations between metaphonological abilities and reading were highly significant for the
kindergarteners. In the tasks involving sensitivity to phonological similarities, correlations were
weak and nonsignificant for the first graders. A principal components analysis shows two
components at first grade: sensitivity to phonological similarities and segmental awareness.
Reading was related only to the latter. The differential performance between prereaders and
readers within the group of kindergarten shows that sensitivity to phonological similarities and
initial isolation of segments takes precedence over alphabetic reading. Segmental awareness,
however, does not develop outside the learning of the alphabetical code as the evidence provided
by results in deleting, counting and reversal tasks suggests. All children who had developed
segmental awareness were able to read but, interestingly enough, some good readers performed
poorly in some of the segmental awareness tasks (i.e. deleting of initial phoneme).

KEY WORDS: Early reading, Levels of phonological awareness, Phonological awareness

INTRODUCTION

The central role of phonological awareness in learning to read and write in


an alphabetic code is becoming widely recognized. Nevertheless, there have
been differences in interpreting its role in reading acquisition. Some authors
(i.e. Bradley & Bryant 1985; Lundberg, Olofsson & Wall 1980) suggest it as
a precursor of reading while others (i.e. Morais, Alegrfa & Content 1987)
claim that, at least some aspect of phonological awareness results from reading
acquisition, or is a concomitant of it (cf. Brady & Shankweiler 1991).
Although the interactiveview propound by Morais et al. (1987) may satisfy
the tenets of the two previously mentioned positions, it seems necessary to
differentiate 'which' form or level of phonological awareness precedes reading
acquisition and 'which' follows it. Thus, making progress in the identifica-
tion of different levels of phonological awareness is a priority goal.
During the last decade many researchers have indicated that different tasks
used to operationalize the concept of phonological awareness may require
different degrees or levels of metaphonological skills (Backman 1983;
Nesdale, Herriman & Tunmer 1984) and studies have tried to specify those

Reading and Writing: An Interdisciplinary Journal 6: 279-298, 1994.


© 1994 Kluwer Academic Publishers. Printed in the Netherlands.
280 M. C A R R I L L O

levels by means of a non mutually exclusive approach, for instance in terms


of the relative difficulty of tasks (Stanovich, Cunningham & Cramer 1984;
Yopp 1988), in terms of their developmental sequence (Perfetti, Beck, Bell
& Hughes 1987) or in relation to the linguistic level of the phonological unit
to take into account (Treiman & Zukowski 1991).
Recently, Morais (1991a) has suggested a taxonomy which includes four
types of phonological awareness differentially related to reading acquisition:
awareness of phonological string (non-analytical level), awareness of sylla-
bles, awareness of phonemes (segmental awareness), and awareness of
phonetic features. Several studies show that awareness of syllable and
awareness of phonological strings (i.e. sensitivity to rhyme) can precede
literacy instruction in many children (Bradley & Bryant 1983; Liberman,
Shankweiler, Fischer & Carter 1984), while segmental awareness seems to
require confrontation with the alphabetical code (Morais et al. 1987).
Goswami & Bryant (1990) have also distinguished different levels of
phonological awareness (syllabic, intra-syllabic and phonemic) with different
relationship to reading. Their model supports a direct connection between
children's sensitivity to rhyme before they read and their later success in
reading. 1 However, many studies have not found that connection (Lundberg,
Frost & Petersen 1988; Stanovich et al. t984; Yopp 1988).
Although the reciprocal relationship between phonemic awareness level
and reading has been well established, some disagreement remains with regard
to the source and the mechanisms underlying its development. While some
investigators suggest that phonemic awareness is a critical precursor of reading
acquisition and that it can be observed among some non-readers (Lundberg,
1991), others maintain that acquisition of phonem!c awareness requires explicit
instruction on the alphabetic code (Morais et al. 1987). It seems possible that
some metaphonological tasks (i.e., categorization, detection and deletion of
segments) can be performed successfully without being able to consciously
represent and manipulate phonemes (Morais 1991a). As a result, the success
in those tasks must be interpreted with caution.
So, some of the problems encountered to establish causal links between
phonological awareness and reading concern the types of different tasks used
as well as the processes used to perform them.
An approach to Sorting out these problems consists of giving the same
group of children several metaphonological tasks measuring different kinds
of metaphonological abilities (Lenchner, Gerber & Routh 1990; Stanovich et
al. 1984; Yopp 1988). This procedure enables the researcher to compare the
extent to which these tasks tap a similar underlying construct of phonolog-
ical awareness as well as its relative predictive power of reading ability.
The present study makes use of that procedure to explore the relations
between different levels of phonological awareness and learning to read in
Spanish language.
D E V E L O P M E N T OF P H O N O L O G I C A L AWARENESS 281

METHOD

Subjects. The children participating in the study (n = 120; 56 males and 64


females) were selected randomly from five kindergarten classes (n = 68) and
four first grade classes (n = 52) in three schools in Murcia (Spain). The vast
majority of the subjects in both groups were from middle class families.
The mean age, when the evaluations were carried out (at the end of school
year) was 5 years and 10 months for kindergarteners (age range of 5;5 to
6;4) and 6 years and 10 months for first graders (range of 6;5 to 7;3). The
mean CGI (Cognitive General Index) on the McCarthy Scales of Children's
Abilities (McCarthy 1972) was 105 for kindergarteners and 106 for first
graders. None of their CGI scores was lower than 85.

Materials and procedure. Ten phonological awareness tasks were individu-


ally administered to the subjects in two or three separate sessions.
Task 1. Sensitivity to rhyme. Six series of four words (CVC) were
composed: In each series three words shared the two final segments (-VC)
and only one shared the vocalic segment (-V-). The first series was used for
training purposes (i.e. pan, dan, real, san). In kindergarten, the procedure
was to compare successively the first word of each series (which was accom-
panied by its picture) against each other. The children had to detect if the
pair of words ended with the s'ame sound (i.e. pan-dan; pan-mal; pan-san).
The score was the number of items successfully answered (score max = 15).
In first grade, the experimenter named the four words of each series and the
children had to detect the only word that did not rhyme with the others (score
max = 5).
Task 2. Sensitivity to alliteration. The stimulus and procedure were similar
to those of task 1. But in this task three of the words in each series shared
the two initials segments (CV-) and the different word shared only the vocalic
segment (-V-; i.e. sol, voz, son, sor). The target segment was the initial one.
Task 3. Sensitivity to rhyme. This task was also similar to task 1, both in
the orthographic pattern of the stimulus and in the procedure. But in this
task, the different word in each series only shared the final consonant ( - - C ;
i.e. pez, vez, tez, coz).
Task 4. Position segment identification. The stimulus consisted for six series
of four words with each series sharing a target segment (different in each
series) at different positions (initial, medial or final; i . e . / a / i n ola, ajo, cae,
mar). The children task was to say the position of the target segment of each
item in the series (i.e. where i s / a / i n 'ola', at the beginning, the middle or
the end?; where is/a/in ' a j o ' . . . ? etc.). The first series was used for training.
The score was the number of items identified successfully (score max = 20).
Task 5. Final segment deleting. Fifteen words, eight of them with three
segments (CVC, CVV, or V-CV) and seven with four or five (CCVC, CV-CV
or CCV-CV) we're the experimental stimulus and another six words were used
for training. The children were supposed to say a non-word by deleting the
282 M. CARRILLO

final segment of each word named by the experimenter (i.e.,/dos/--->/do/).


The score was also the number of correct responses (max = 15).
Task 6. Initial segment deleting. The stimulus and procedure were similar
to those of task 5. But in this task, the segment to omit was the initial one
(i.e.,/pan/~/an/).
Task 7. Initial segment isolation. The experimental stimulus consisted of
ten words (CVC) while another three words were used for training purposes.
The task involved the production of only the first segment of each word
previously pronounced by the experimenter ( i . e . , / m a r / ~ / m / ) .
Task 8. Final segment isolation. Both stimulus and procedure were similar
to those of task 7. But in this task the segment to be produced was the last
in each word ( i . e . , / s o l / ~ / I / ) .
Task 9. Total segmentation. Ten words (three CV, two V-CV and CV-CV
and one CVC, VC-CV and V-CCV) were included in this task (in addition
to four items for training). The children should, successively, figure out all
segments of each word and put on the board the same number of chips as
segments (i.e.,/mar/--~/m//a//r/).
Task 10. Reversal segments. Ten words (three CV three VC, two CVC and
two V-CV) comprised the experimental stimulus and another four similar
words were used for training. The children were to say a real word or a non-
word by reversing the order of segments in each word (i.e., ~sol~ --~ ~los~). The
maximum score in the last six tasks was ten for each of them. Moreover, in
all tasks, the correct answer was provided only for the training items.

In kindergarten, the tasks were administered in three successive days: tasks


1, 4 and 5 during the first session, tasks 2, 6, 7 and 8 during the second session
and tasks 3, 9 and 10 during the third session. In first grade tasks 1, 2, 3, 4,
7 and 8 were administered in one session and tasks 5, 6, 9, and 10 in another.
In addition, the children were evaluated in knowledge of letter sound (L1) and
in their ability to decode syllables and words (L2). For kindergarteners L1
included twenty seven letters and L2 included thirty five syllables (20 CV,
5 VC, 5 CVC, 3 CCV and 2 CCVC), 2 four monosyllabic words (CVC), thirty-
four bisyllabic items (VCV, CVCV, CCVCV, CVCCV) and seven trisyltabic
words. Only a few words (between three and ten depending on the class-
room) of the 80 in the test were included in reading material of the
kindergarteners. In first grade, L1 included sixteen letters and L2 included
thirty six syllables, twenty four bisyllabic and fifteen trisyllabic words with
orthographic patterns as in the kindergarten test. Both the metaphonological
tasks and the reading tests were administered at the end of the school year.

RESULTS

Development of phonological awareness. In order to facilitate direct com-


parison of different metaphonological tasks, scores were turned into success
DEVELOPMENT OF PHONOLOGICAL AWARENESS 283

proportions. M o r e o v e r a correction for guessing 3 was used for tasks 1, 2, 3


and 4. In Table 1 descriptive statistics (mean, standard deviation and median)
o f metaphonological tasks are presented for each grade (kindergarten and
first grade). Figure 1 shows differences between both grades in mean success
proportions for each task.
At kindergarten, initial deleting, final isolation, counting and reversal
segments were the m o s t difficult tasks, the mean proportions being poor
(between 0.10 and 0.39); besides, the smaller values of medians (between 0.00
and 0.25) suggest that the majority o f the kindergarteners were unable to do
those tasks. However, at 1st grade, the mean proportions were similar in alt
tasks, although a close look at medians revealed that the most difficult task
was initial segment deleting for a great numbers o f subjects (cf. Bruce 1964;
Stanovich et al. 1984; Yopp 1988).
A two-way A N O V A was c a r d e d out to compare kindergarten and 1 st grade
at the different metaphonotogical tasks. The effects for grade and task were
both significant (Fl,11so = 276.55; p < 0.001 and Fg, Hs0 = 7.81; p < 0.001
respectivily) as well as their interaction (Fg, zls0 = 7.97; p < 0.001). Post hoc
comparisons (by means of the Bonferroni procedure) indicated that kinder-
garteners scored lower than the first graders in all tasks and that differences
were stronger in initial deleting, counting and reversing segments tasks (p <
0.001). Smaller and no significant differences appeared in rhyming and

Table 1. Descriptive statistics per grade for metaphonological tasks (in proportions of correct
responses) and reading (in direct scores)

Task Kindergarten (n = 68) First grade (n = 52)

Mean SD Mdn Mean SD Mdn

1. Rhyme: Final consonant


diff. 0.52 0.37 0.53 0.62 0.36 0.73
2. Alliteration: Init.
consonant diff. 0.52 0.34 0.53 0.56 0.35 0.73
3. Rhyme: Medial vowel diff. 0.49 0.34 0.47 0.70 0.31 0.73
4. Position segment
identification 0.45 0.40 0.44 0.80 0.30 1.00
5. Final segment deletion 0.45 0.26 0.40 0.77 0.24 0.87
6. Initial segment deletion 0.14 0.23 0.00 0°64 0.30 0.67
7. Initial segment isolation 0.48 0.42 0.50 0.77 0.35 1.00
8. Final segment isolation 0.38 0.40 0.25 0.77 0.37 1.00
9. Total segmentation 0.39 0.41 0.20 0.80 0.24 0.90
10. Reversal segments 0.10 0.25 0.00 0.71 0.28 0.80
L1 Letter-sound knowledge~ 19.09 7.86 23.00 14.94 1.50 15.50
L1 Syllables and words
decodingb 36.69 30.06 38.00 58.56 14.75 64.00

a For kindergarten, score max = 27. For first grade, score max = 16.
b For kindergarten, score max = 80. For first grade, score max = 74.
284 M. C A R R I L L O

~ . -t////itt//~//. -/////a
Reversal

Tot. segment ~ ~ l / / / l i / t , , ) ' / / / / / l / / . a


Final isolat ~ . ' l t / H l l / , ' i . ) l l / / / i l A
Inil. isolat
Inil. delet, ~.'//li/zl~ll////H/I//I.'/12dJ/A
Final dele t. L~liZili/.~/////zl/lt//,//.','l.'l/,'lz.'/.'/.'.'/.'//.','/.'/.'.'~'~////////A
Pos identif,
F~yme medial V/ii/fi/////t/l/X////i///i//~.,//.c///x///x////fa/g~
I

I .
Alliteration r f l l l l l l l ~ l l l l l l l l ~ $ 71t11iiiit1111111111t~

Rhyme final [] FIRST


[~ .i___ , i_ ,i ....... [] KINDER
0.0 0.2 0.4 0.6 0.8 1.0

Moan Prooortion of Correct Responses

Figure 1. Mean proportions of correct responses per grade (Kindergarten and First grade) for
each metaphonologicaltask.

alliteration (tasks 1, 2 and 3). In the other tasks (4, 5, 7 and 8) differences
were also strong (p < 0.001). Comparisons between the ten metaphono-
logical tasks for each grade indicated significant differences in both initial
deleting and reversal tasks and in the remaining ones (p < 0.005) for kinder-
garten. No other comparisons were significant either for kindergarten or for
first grade.

Intercorrelations between metaphonologicat tasks. Interrelationships among


metaphonological tasks at each grade are presented in Table 2. In kindergarten,
most correlations are statistically significant (p < 0.001). But, surprisingly,
one of the lower correlations (r = 0.31; p < 0.01) was between tasks 1 and 3,
both of rhyme detection. At first grade tasks 1, 2 and 3 showed moderate
relationship with each other but presented weak correlations (non significant
at p < 0.001) with the other tasks. Correlations among the last seven tasks
were moderate to strong. All of them, except for the correlations between the
final segment deleting task and both segment position identification and total
sound segmentation tasks, were larger than 0.43 (p < 0.001).

Principal components of the metaphonological abilities. Results of a principal


component analysis made on the data of all ten tasks in each grade separately
indicated that (see Table 3): (1) In kindergarten the two rotated components
(with eigen values > 1) accounted for 38.16 and 27.26% respectively of the
total variance. In the first component, the tasks that had high loadings were
related to sensitivity to phonological similarities (tasks t and 2) and identi-
fication or isolation of segments (tasks 4, 7 and 8). The tasks of initial deletion
(task 6) and reversal (task 10) had high loading in the second component.
T a b l e 2. Intercorrelations of metapbonological tasks, reading and Cognitive General Index (CGI) for Kindergarten* (above the diagonal) and First
grade** (below the diagonal)

Task 1 2 3 4 5 6 7 8 9 10 L2
©
1. Ryme: Final consonant
diff. - 0.44 0.31 0.68 0.48 0.45 0.58 0.60 0.53 0.40 0.56 t'n
2. Alliteration: Initial Z
consonant diff. 0.55 - 0.40 0.42 0.28 0.37 0.43 0.55 0.40 0.23 0.40
O
3. Ryme: Medial vowel
diff. 0.42 0.47 - 0.40 0.35 0.41 0.45 0.41 0.50 0.42 0.51
4. Position segment 0
identification 0.37 0.39 0.29 - 0.62 0.44 0.59 0.58 0.68 0.35 0.71 Z
5. Final segment deletion 0.31 0.29 0.20 0.40 - 0.51 0.43 0.42 0.49 0.50 0.45 0
6. Initial segment deletion 0.20 0.27 0.13 0.52 0.60 - 0.49 0.59 0.53 0.83 0.54 0
7. Initial segment isolation 0.30 0.36 0.24 0.72 0.49 0.51 - 0.82 0.60 0.28 0.58 0
8. Final segment isolation 0.26 0.32 0.16 0.66 0.52 0.55 0.95 - 0.54 0.36 0.61
9. Total segmentation 0.05 0.32 0.09 0.57 0.31 0.56 0.57 0.6! - 0.47 0°89
10. Reversal segments 0.37 0.40 0.21 0.79 0.54 0.67 0.67 0.66 0.66 - 0.51
READING (L2) 0.15 0.14 0.10 0.56 0.55 0.52 0.59 0.69 0.51 0.68 -

C G I - Kindergarten 0.13 0.03 -0.04 0.02 0.14 0.07 -0.17 -0.10 0.09 -0.09 0.08
t'rl
CGI - First grade 0.37 0.16 0.47 0.31 0.07 0.10 0.14 0.08 0.16 0.24 0.07

* For r >- 0.40, p < 0.001. r~


** For r > 0.43, p < 0.001.

t~
286 M. CARRILLO

Table 3. Principal component analysis of the performance in the metaphonoiogicaltasks per


grade. Rotate loadings on components(Eigen value > 1)

Task Kindergarten Task First grade

Comp. t Comp.2 Comp. 1 Comp.2

7. Initial isolation 0.835 0.209 7. 0.847 0.217


8. Final isolation 0.820 0.270 8. 0.872 0.136
4. Pos. identification 0.751 0.335 4. 0.774 0.315
1. Rhymefinal 0.723 0.302 5. 0,614 0.254
2. Alliteration 0.684 0.104 6. 0.773 0.081
3. Rhymemedial 0.444 0.443 9. 0.791 -0.019
5. Finaldeletion 0.418 0.596 10. 0.846 0.255
9. Total segmentation 0.643 0.465 1. 0.154 0.812
6. Initialdeletion 0.313 0.850 3. 0.054 0.784
10. Reversal 0.093 0.952 2. 0.273 0~764
% of total variance 38.156 27.265 44.928 21.582

For the other tasks (tasks 3, 5 and 9) the loadings were similarly distributed
between the two components; (2) In the first grade, the principal component
accounted for 44.93% of total variance. In this component all tasks, except
for those related to sensitivity to phonological similarities, had strong toadings.
On the contrary, only these tasks (1, 2 and 3) were included in the second
component (21.58% of total variance explained).

Relations between metaphonologicaI abilities and reading. Many kinder-


garteners were able to read (cf. Table 1): half of them (34 children)
succeeded in reading, at least, thirty eight syllables and words. Thirty of these
early readers came from three classes in which a phonics method was used,
In the other two kindergarten groups participating in the study, the method
was whole-word, although one of them also included letter-sound instruc-
tion. In this group, four out of fourteen children scored over median value
(Mdn = 38). No children of the other whole-word class reached the median
value in reading (score max = 8). Reading instruction in all first grade
classrooms was phonics method.
For kindergarten, all correlations between the metaphonological tasks and
reading (L2) (cf. Table 2) were at least of 0.40 (p < 0.001) in spite of a floor
effect in some of them (i.e. tasks 6 and 10). For first grade, however, rhyming
and alliteration tasks did not present any correlation with reading: these tasks
were poor predictors of reading achievement and were only weakly related
to the other phonological awareness measures. In accordance with these
results, a second principal component analysis, performed after adding scores
in reading to scores in metaphonological tasks, indicated that reading ability
was associated with the two components obtained for kindergarten. For first
grade, however, reading was only associated with the component which did
not include any rhyming tasks (see Table 4).
D E V E L O P M E N T OF P H O N O L O G I C A L A W A R E N E S S 287

Table 4. Principal component analysis of the performance in the metaphonological tasks and
reading per grade, Rotated loadings on components (Eigen value > 1)

Task Kindergarten Task First grade

Comp. 1 Comp. 2 Comp, 1 Comp. 2

7. Initial isolation 0.835 0,200 7. 0.830 0,242


8. Final isolation 0.811 0.262 8. 0.873 0,150
4. Pos. identification 0.728 0.349 10. 0.845 0.268
I. Rhyme final 0,716 0,294 Reading 0.816 -0,010
2, Alliteration 0,668 0,094 9, 0,765 0,020
Reading 0.691 0,492 4, 0,759 0,337
3, Rhyme medial 0.448 0,544 6, 0,755 0,106
9, Total segmentation 0,679 0,477 5. 0.630 0.244
5. Final deletion 0,406 0,586 1, 0.146 0.808
6. Initial deletion 0.306 0,844 2, 0,238 0,787
10. Reversal 0.095 0,951 3. 0,051 0.775
% of total variance 38.936 26.707 45.807 20,101

Scatterplots of the relations between the performance in metaphonological


tasks and reading allow us to examine individual cases.
Distributions are not the same for all tasks. In those tasks that involve
sensitivity to sound similarities (tasks 1, 2, 3) as well as in those that involve
detection or isolation of segments (task 4, 7 and 8) some kindergarteners
that reached at least 0.60 of correct responses scored low in reading; most of
them were from the two whole-word classes, but, specifically in detection and
isolation tasks, all of them came from the whole-word group with letter-sound
instruction. Besides, in final segment deleting many prereaders were suc-
cessful. Also, some first graders scored high in tasks 1, 2, 3, 4 and 5 but
were relatively poor in reading. However, in initial deleting and reversal
(Figure 2) no poor reader of first grade excedeed 0.60 of success and only
two of them excedeed that level in the total segmentation task (Figure 3).
In contrast, in all tasks there were some good readers that scored
relatively low. This situation was more frequent for initial deletion, final
isolation and reversal tasks, specially in kindergarten where only some of the
best readers were successful (see Figure 2 for tasks 6 and 10). Although this
instance was less extreme for the total segmentation (Figure 3), even in this
task six kindergarteners that scored high on the decoding test (between 0.44
and 0.62) were very poor (between 0.00 and 0.10), as was the case with four
first graders good readers (between 0.74 and 0.94 on decoding test), who
scored no more than 0.40.
So, although scatterplots in some studies (i.e. Tunmer & Nesdate 1985)
have shown that many children performed well on phoneme segmentation
tasks but poorly on decoding, and no children performed poorly on phoneme
segmentation but well on decoding, our results show an opposite pattern in
the more specifically segmental tasks and a different pattern for sensitivity
tO
Oo
Oo

Kindergarten First Grade

1,1 I ~ I I - ' ~ - ' ~ T ~ 1,1 T r T T'-- 1 '--~]

o~
o
0,9 0,9

09 o o
e) -

r"
0,7 0.7
o o o ; / y °~ o
o o /./ ~® .
< d3
0,5 0,5
C o o
CO

09 0,3 0.3 o o o .
cO o o o /o 0
o o
o
01 0.1
TASK10
O TASK6
0.1 0.3 0.5 0.7 0,9 1.1 0.1 0,3 0.5 0,7 0.9 1.
READING READING

Figure 2. Individual results for scores ]n task 6 (initial deletion) and task 10 (reversal) in relation to reading (Kindergarten and First grade).
Kindergarten First Grade

1.1 - - I I I I I
~7
1.1 I I I i I
<
O O O 000 O O:DnnO

0 0 CIDO O] - 0
0.9 0.9 O O (3:131::130 -

OOQ ~3
0 0 0 0000

c 1"7 O 0 CI 0
.O_ 0.7 0.7 0 0 O0 0
0
c 0 o °

E 0.5 0.5 o
O0 0 Cl 0 0 C~ z
69
o
O
0.3 0
I-- 0.3 0
E
0,1 0 0
0.1
nno 0CI¢7 0 0 0 0
o

I ____1 I I I t~
0,1 0.3 0.5 0.7 0.9 1.1 Z
0.1 0.3 0.5 0.7 0.9 1.1
READING READING

Figure 3. Individual results for scores in task 9 (total segmentation) in relation to reading (Kindergarten and First grade). t~
290 M. CARRILLO

to sound similarities and detection of segments, in which there were both some
good readers with (relatively) low metaphonological abilities and some poor
readers with (relatively) high metaphonological level.

Differential analysis for reading level. The occurrence of strong differences


in reading ability, specially in kindergarten, makes necessary a further analysis
to study the development of metaphonological abilities both in prereaders and
early readers. Similarly, in first grade performance between good and poor
readers should be compared in order to explore the metaphonological level
in which the differences appear.
Forty four kindergarteners (67.8% o f t h e total sample) were able to apply
the alphabetical principle to syllables and word decoding, as indicated by their
scores in the reading test, which were at least twelve items correctly read
(15% of items included). The other twenty four kindergarteners were con-
sidered prereaders since they identified only a few familiar items (L2 < 15%).
Almost all of these came from the two whole-word classes: fifteen attended
the strict whole-word class and five came from the other class where letter-
sound correspondences and a limited number of sight words had been intro-
duced.
Table 5 (and Figure 4) shows the relative size of the differences in per-
formance between prereaders and early readers at each metaphonological task.
For all tasks, early readers exceeded at least 15% but differences were more
remarkable for initial deletion, final isolation, total segmentation and reversal
tasks, in which prereaders scored very low. Kindergarten early readers reached
high scores in most of the tasks, only at initial deleting and reversal segments

Reversal
TO{. t~l"lla£1| L.II/i/ii/;'//,6,//ilIIII/////////////////////////A
Final isolat ~ / / ...............
Inii isolat ~ . - z . ~ 4 :..:;.
.~.,> "///'////"/"/1"'////////'4
03 Inil 0eleL LIIIIIlilT///I/k
~//////H/~fM//////U//////////.H//H/~
O3

POS iOerilil. L/i/i////is//////.,'/////////////////////////////////J

Rhyme meO~l L'/I///S//////I/////M/iZ////HH/HH/.~Z.'////~


Allilera lion L//////////////////w///////////.~///////////////A
Rhyme lina! L////Iz//I//~¢/II×.//////////////A
E3 EARLYR
t I I I I Q PREREAD
0.0 0.2 0.4 0.6 0.8 1.0
Mean ProDorlion of Correct Responses

Figure 4. Mean proportions of correct responses per reading level at Kindergarten: prereaders
group (PREREAD) and early readers group (EARLYR) for each metaphonological task.
DEVELOPMENT OF PHONOLOGICAL AWARENESS 291

Table 5. Differential performance for reading level per grade: mean, standard deviation ( ) -
in proportion of correct responses - and comparison probabilities (by Bonferroni adjustment)

Task Kindergarten First grade

Pre- Early p Average and Good p


readers readers poor readers readers

1. Rhyme: final 0.33 0.62 0,037 0.57 0.65 1.000


consonant diff. (0.30) (0.36) (0.36) (0.36)
2. Alliteration: Initial 0.36 0.61 0,238 0.48 0.62 1.000
consonant diff. (0.34) (0.31) (0.34) (0.34)
3. Rhyme: Medial 0.32 0.58 0.237 0.65 0.74 1.000
vowel diff. (0.27) (0.34) (0.32) (0.31)
4. Position segment 0.13 0.63 0.000 0.64 0.93 0.078
identification (0.20) (0.38) (0.36) (0.I 1)
5. Final segment 0.35 0.50 1,000 0.65 0.86 1.000
deletion (0.21) (0.27) (0.27) (0.18)
6. Initial segment 0.03 0.20 1.000 0.50 0.75 0.398
deletion (0.07) (0.26) (0.28) (0.28)
7. Initial segment 0.18 0.65 0.000 0.59 0.91 0.018
isolation (0.28) (0.40) (0,39) (0.23)
8. Final segment 0.07 0.55 0.000 0.56 0.92 0.002
isolation (0.16) (0.40) (0.42) (0.23)
9. Total 0.03 0.59 0.000 0.69 0.88 1.000
segmentation (0.08) (0.39) (0.25) (0.20)
10. Reversal 0.00 0.15 1.000 0,51 0.87 0.002
segments (0.00) (0.30) (0.29) (0.14)
L1 Letter-sound 0.39 0.88 0.89 0.97
knowledge (0.22) (0. t5) (0.12) (0.05)
L2 Syllables and 0.04 0.69 0.64 0.91
word decoding (0.04) (0.26) (0.22) (0.04)

Prereaders: Syllables and word decoding test < 0.15 (12 items).
Early readers: Syllables and word decoding test ~ 0.15.
Average and poor readers: Syllables and word decoding test < 0.85 (63 items).
Good readers: Syllables and word decoding test ~ 0.85.

the mean scores were lower than 0.50. As a matter of fact in these tasks only
the best readers were successful and most early readers scored below 0.30 of
correct responses (see Figure 2).
In a two-way analysis of variance with reading level (prereaders vs. early
readers) and task (10 tasks) as factors, the main effect of reading level was
significant (Fl,g~0 = 177,08, p < 0.001) as well as the effect of task (F9,660 =
13.73, p < 0.001). Post-hoc comparisons (by Bonferroni procedure) indicated
that, for prereader level, only differences between task 10 (reversal) and tasks
1, 2, 3 (rhyme and alliteration) and 5 (final deletion) as well as between task
2 (alliteration) and tasks 6 and 9 (initial deletion and total segmentation
respectively) were significant (p < 0.05), while in early reader level, highly
significant differences (p < 0.001) occur between initial deletion and reversal
292 M. C A R R I L L O

with the remaining ones, Comparisons between two reading levels of each
task (Table 5) show highly significant differences (p < 0.001) for tasks 4, 7,
8 and 9 (detection and isolation of segments).
At first grade, performance in the reading test was very high (mean = 0.79
and median = 0.86). The good readers (thirty out of the 52 children) were able
to read at least 85% of the items included (mean = 0.91) while the remaining
children (n = 22) were considered average and poor readers (mean = 0.64).
Their scores in the metaphonological tasks were as follows (Table 5 and Figure
5): In tasks 1, 2 and 3, good readers exceeded the average and poor readers
group between 8% and 14%, but in the other tasks differences were higher
(between 19 % and 36%) and specially dramatic in isolation segment tasks,
in which even kindergarten early readers were a little better than the average
and poor readers group.

.Or(if$~i1~ ..........;/t/~
TOt, se~"nenl L////I/I////llll~t/I//.+'///i///I/////i~d
~,~..~,
........ ~,,:~..,,,.,;~i~a~.:~l
Fir,~l isol~t ~/////////~//,'~'/'~'///~/i///////~//'//////~//////H//////~////////1j///////A
IrI|.i$ol~ll ~ll.,/#//lfi/i/:llll/~l.L~
Inil delet. ~Z.'."/H//,'Z/i,'Z///I////.'~//H/////H/////////~

~//x/////Xl//ll~Xll/////////////f/~/~
~ W~I Lili//il#11ill/ii/l/#~ililililltlllllilll~
Atlltera l i e ~ltl/llll/ltlllllllllllltlllllZ.lllltlll/lltl~-illiA
Nine
~ll¢lllltllt)vTiilillltllllllllillitlllllllllllltl~
llrlel ............................... ::" ~+.........~ [] GOODR
i I ~ I N AVPOOR
0,0 0,2 0.4 0.6 0.8 1,0

Mean Proportion of Correct Responses

Figure 5. Mean proportions of correct responses per reading level at First grade: average and
poor readers group (AVPOOR) and good readers group (GOODR) for each metaphonological
task.

As in kindergarten, a two-way analysis of variance confirmed significant


main effects both for reading level (average and poor readers vs. good readers)
and task (Fi,50o = 80.04, p < 0.001 and F9.50o= 3.90, p < 0.001 respectively),
Correspondent post-hoc comparisons between two reading levels at each task
indicated significant differences (p < 0.005) for final isolation and reversal
tasks.

DISCUSSION

The goal of this study is the identification of different levels of phonological


DEVELOPMENT OF PHONOLOGICAL AWARENESS 293

awareness as well as the differentiation of their relationship to reading level


in Spanish children. To this end, a comparison was made in different
metaphonological tasks both within kindergarten and first grade groups and
between these two groups. Moreover, differences found were put in relation
to success level in reading.
The differences observed in mean scores for each task at kindergarten
confirmed results of previous studies in which early readers were given a
battery of phonological awareness tasks (Stanovich et al. 1984; Yopp 1988).
Those results indicate that performance varies greatly across different tasks,
with rhyme and alliteration being the easiest and deletion and reversal the
most difficult ones. However, our results do not show such differences at
first grade since at this level mean scores were similar for all tasks, regard-
less of their cognitive demands.
On the other hand, differences observed both in performance and rate of
development between some groups of tasks suggest that they tap distinct kinds
of phonological awareness (see Morals 1991b): sensitivity to phonological
similarities as well as ability to detect and isolate initial segments (tasks 1, 2
and 3, and tasks 4 and 7 respectively), which develop early in kindergarten,
while segmental awareness (tasks 6 and 10 and probably tasks 8 and 9) appears
later on. With regard to performance in final isolation and total segmentation
(tasks 8 and 9) mean scores suggest (in comparison with task 7) that in order
to isolate non-initial segments (i.e., final and medial segments) some com-
petence, which is not necessary to isolate the initial segment, is required. That
type of competence probably needs conscious representations of phonolog-
ical segments (Morals 1991b). On the contrary, it is believed that the final
segment deletion task (task 5) may be performed by kindergarteners without
a explicit segmental analysis, for instance, 'to monitor one's articulation of a
utterance and stop just before the last articulatory gesture' (Content, Kolinsky,
Morals & Bertelson 1986: 67). Similarly, the isolation of initial segment may
be performed stopping just after producing the first articulatory gestures.
Besides, the differential rate of development from kindergarten to first
grade for tasks 1, 2 and 3, in relation to the other tasks indicates that some
developmental variable (i.e. the confrontation with the alphabetical code)
differentially affects analytic and non-analytic metaphonological abilities.
The intercorrelations between tasks as well as the principal components
found for each grade also confirm the existence of the two forms of
phonological awareness: A holistic and an analytic one, which are perfectly
differentiated when metaphonotogicat development has reached a good level
(at the end of first grade in our study). A similar result was reported by
Stanovich et al. (1984) and Valtin (1984) who found that rhyming tasks were
excluded from the main factor which accounted for performance in the
remaining tasks. However, they justified that exclusion on the basis of a ceiling
effect in the rhyming tasks, a fact which did not occur in our study.
Moreover, performance at kindergarten suggests a metaphonological
capacity underlying early development of the two phonological awareness
294 M. CARRILLO

forms, as well as a different capacity underlying more advanced analytic


ability. Getting this high level of analytic phonological awareness is probably
dependent on the acquisition of conscious representations of phonological
segments. That form of phonological awareness (Morais et al. 1987) enables
the subjects to perform successfully different kinds of analytic metaphono-
logical tasks, regardless the cognitive operations implied. Many of our first
graders have probably developed that level of segmental awareness just as it
is indicated by their similar mean scores in all tasks administered.
The relationships between the development of the different forms of
metaphonological abilities and reading acquisition provide some important
data to support distinct kinds and levels of phonological awareness.
The most relevant data are the lack of relations between holistic phono-
logical awareness and reading ability at first grade. The second principal
component analysis performed, including reading, as well as results of
differential performance between first grade good readers and first grade
average and poor readers, confirm that the acquisition of relatively high
decoding ability is associated with analytic awareness but is independent from
holistic phonological awareness. Nevertheless, results of the same analysis for
kindergarteners indicated that the most rudimentary phonological decoding
ability is associated with a more basic and general metaphonological capacity
which underlies both holistic and analytic phonological awareness.
On the basis of differential performance between prereaders and early
readers at different tasks a more complete view of the development of
phonological awareness may be obtained. In the prereaders group, although
significant differences only occurred between some tasks, mean scores suggest
that sensitivity to phonological ~milarities and initial isolation of segment
begin to develop before learning to read, probably as a result of a basic
metaphonological capacity. However, there is no similar indication with
respect to the development of segmental awareness. Together with initial
acquisition of alphabetic code, early readers begin to develop basic segmental
abilities, as indicated by similar performance in all tasks except in those
involving a more advanced level of segmental awareness. In addition, some
kindergarteners with high reading ability reach an important development of
segmental awareness.
On the other hand, the fact that the highest differences between prereaders
and early readers occur in tasks involving detection and isolation of segments
suggests that acquisition of basic segmental abilities is the critical form of
metaphonologicat development at the beginning of reading acquisition. In
support of such critical role for basic segmental abilities, it must be pointed
out that differences between average and poor readers group and the good
readers group at first grade were significant for segment isolation tasks, with
mean scores from the former group being less than 0.60. In addition, an
examination of the performance of the six poorest readers (L2 < 0.50) shows
very low mean scores in those tasks (0.22 and 0.10 for tasks 7 and 8 respec-
D E V E L O P M E N T OF P H O N O L O G I C A L A W A R E N E S S 295

tively) even as low as the more genuine segmental awareness tasks (i.e., initial
deletion and reversal).
In order to account for the source of acquisition of segmental awareness,
the individual performance of prereaders in segment isolation tasks was
examined, since, in accordance with previous conclusions, this ability proved
to be critical for that acquisition. Scatterplots show that ~,;ome prereaders
reached a high level in those tasks, a fact which suggests that basic analytic
phonological ability may develop independently from reading acquisition.
However, it should be remarked that these prereaders (all of them) proceeded
from the whole-word class that have received letter-sound instruction, while
no prereader child from the other whole-word class reached a similar level in
segment isolation tasks. Therefore, results indicate that initial development of
analytic awareness was elicited by explicit instruction in grapheme-phoneme
correspondences and they are in agreement with the notion that segmental
awareness develops only through explicit phonic instruction (Alegrfa, Pignot
& Morais 1982; Alegrfa, Morais & D'Alimonte, in prep.; Morais, Cary,
Alegrfa & Bertelson 1979; Read, Zhang, Nie & Ding 1986).
Finally, with respect to a more advanced level of segmental awareness, as
revealed by initial deleting and reversal tasks, differential performance in
accordance with reading level (Table 5) indicated that the great majority of
early readers does not develop this level of phonological awareness since only
some of the best ones reached scores higher than 0.40 in those tasks. However,
the average and poor readers group from first grade reached similar mean
scores for all metaphonological tasks. This suggests that many first graders
have a good level of segmental awareness available, although the six poorest
readers (L2 < 50%) were almost as poor as the early reader group of
kindergarten in those tasks. Since both early readers from kindergarten and
average readers from first grade showed high decoding ability their differ-
ences in segmental awareness suggest that, in order to acquire a full
development of phonological awareness, some amount of training or experi-
ence in reading, not provided in kindergarten, may be required.
As well as for isolating tasks, individual performance for initial deletion
and reversal segments in relation to reading ability was examined: unlike other
tasks, in these two in particular, all good segmentators (both kindergarteners
and first graders) were good readers, and, in accordance with early conclu-
sions, many kindergarteners good readers were poor segmentators. But the
more surprising result was to find some first grade good readers who failed
those segmental tasks. This was the case with four children for deletion and
two for reversal tasks who, in spite of answering correctly at least 80% of
the items in the reading test, scored no more than .20 in the tasks. However,
only two out of the four children who failed the deleting task were also poor
in the reversal task, the other two children scoring 0.90.
So, the fact that many of those good readers - but poor segmentators -
performed poorly in some segmental task but fairly well in another suggests
296 M. C A R R I L L O

that the breakdown may be due to insufficient trials for training together with
the lack of corrective feed-back in experimental trials.
In conclusion, although the correlational data do not allow us to identify
what kind of relation there is between different forms of phonological
awareness and reading (i.e. predictive or concomitant), the examination of
performance in prereaders suggests that awareness of phonological similari-
ties and basic segmental abilities (isolation of initial segments) are precur-
sors. Likewise, the examination of performance in early readers suggests that
segmental awareness is a concomitant. However, the high reliance on
knowledge of letter-sound for the development of basic segmental abilities
and the high improvement on these abilities as reading acquisition develops,
suggests that influences may also occur in the opposite direction. On the
other hand, the very high level reached by first grade good readers in analytic
phonological awareness as well as the lack of improvement on holistic
phonological awareness in relation to average and poor readers, allow us to
emphasize the interactive nature of segmental awareness but to play down its
importance as far as nonsegmentat awareness is concerned.
Segmental awareness appears to be critical for success in reading: all
children who developed segmental awareness were able to read. Nevertheless,
the fact that some good readers continue to perform poorly in some segmental
awareness tasks could be interpreted at least in two ways: It is possible that
a high level of this analytic ability is not a necessary condition to read, or
alternatively, it might be that those tasks were not adequately understood by
some children.

ACKNOWLEDGEMENT

Partially supported by the Centro de Investigaci6n, Documentaci6n y


Evaluaci6n. C.I.D.E. of Ministerio de Educaci6n y Ciencia of Spanish State.
The author would like to thank Julio Roca for his help in preparing the English
version of the text.

NOTES

t. Specifically, they claim that sensitivity to rhyme makes a causal contribution to reading,
because it is the basis to categorise words which share a common final spelling sequence,
and so, it allows using analogies in order to read new words.
2. Eleven of the syllables included were real words, but few of them were common.
3. In tasks 1, 2 and 3 the probability of chance was 0.50, for each item, in kindergarten and
0.25, for each series, in first grade. In task 4 that probability was 0.33 for each item in both
grades.
DEVELOPMENT OF PHONOLOGICAL AWARENESS 297

REFERENCES

Alegrfa, J., Pignot, E. & Morais, J. (1982). Phonetic analysis of speech and memory codes in
beginning readers, Memory & Cognition 10: 451-456.
Alegrfa, J., Morais J. & D'Alimonte, G. (in prep.). The development of speech segmentation
abilities and reading acquisition in a whole word setting.
Backman, J. (1983). The role of psycholinguistic skills in reading acquisition: A look at early
readers, Reading Research Quarterly 18: 466-479.
Bradley, L. & Bryant, P. E. (1983). Categorizing sounds and learning to read: A causal
connection, Nature 301: 419-421.
Bradley, L. & Bryant, P. E. (1985). Rhyme and Reason in Reading and Spelling. I.A.R.L.D.
Monographs, No. 1. Ann Arbor: University of Michigan Press.
Brady, S. A. & Shankweiler, D. P. (1991). Phonological Processes in Literacy. Hillsdale, NJ:
Lawrence Erlbaum Associates.
Content, A., Kolinsky, R., Morals, J. & Bertelson, P. (1986). Phonetic segmentation in
prereaders: Effect of corrective information, Journal of Experimental Child Psychology 42:
49-72.
Goswami, U. & Bryant, P. (1990). Phonological Skills and Learning to Read. Hillsdal, NJ:
Lawrence Erlbaum Associates,
Lenchner, L., Gerber, M. M, & Routh, D. K. (1990). Phonological awareness tasks as
predictors of decoding ability: Beyond segmentation, Journal of Learning Disabilities 23:
240-247.
Liberman, I. Y., Shankweiler, D., Fisher, F. & Carter, B. (1974). Explicit syllable and
phoneme segmentation in the young child, Journal of Experimental Child Psychology 18:
201-212.
Lundberg, I. (1991). Phonemic Awareness can be developed without reading instruction. In:
S. A. Brady & D. P. Shankweiler (eds.), Phonological Processes in Literacy (pp. 47-53).
Hillsdale, NJ: Lawrence Erlbaum Associates.
Lundberg, I., Frost, J. & Petersen, O. (1988). Effects of an extensive program for stimulating
phonological awareness in preschool children, Reading Research Quarterly 23: 263-
284.
Lundberg, I., Olofsson, A. and Wall, S. (1980). Reading and spelling skills in the first school
year predicted from phonemic awareness skills in kindergarten, Scandinavian Journal of
Psychology 21: 159-173.
Morals, J. (199Ia). Constraints on the development of phonemic awareness. In: S. A. Brady &
D. P. Shankweiter (eds.), Phonological Processes in Literacy (pp. 5-27). Hillsdale, NJ:
Lawrence Erlbaum Associates.
Morals, J. (199Ib). Phonological awareness: A bridge between language and literacy. In: D. J.
Sawyer & B. J. Fox (eds.), Phonological Awareness in Reading (pp. 31-71). New York:
Springer Verlag.
Morals, J., Alegria, J. & Content, A. (I987). The relationship between segmental analysis and
alphabetic literacy: An interactive view, Cahiers de Psychologie Cognitive 7: 415-438.
Morais, J., Cary, L., Alegrfa, J. & Bertelson, P. (1979). Does awareness of speech as a sequence
of phones arise spontaneously? Cognition 7: 323-331.
Nesdate~ A. R., Herriman, M. L. & Tunmer, W. E. (1984). Phonological awareness in children.
In: W. E. Tunmer, C. Pratt & M. L. Herriman (eds.), Metalinguistic Awareness in Children:
Theo~Te, Research and Implications (pp. 56-72). New York: Springer Verlag.
Perfetti, C. A., Beck, I., Bell, L. C. & Hughes, C. (1987). Phonemic knowledge and learning to
read are reciprocal: A longitudinal study of first grade children, Merrill-Palmer Quarterly
33: 283-319.
Read, C., Zhang, Y., Nie, H. & Ding, B. (1986). The ability to manipulate speech sounds depends
on knowing alphabetic writing, Cognition 24:31-44.
298 M. CARRILLO

Stanovich, K., Cunninghan, A. E. & Cramer, B. B. (1984). Assessing phonological awareness


in kindergarten children: Issues of task comparability, Journal of Experimental Child
Psychology 38: 175-190.
Treiman, R. & Zukowski, A. (1991). Levels of phonological awareness. In: S. A. Brady & D.
P. Shankweiler (eds.), Phonological Awareness in Literacy (pp. 67-83). Hillsdale, NJ:
Lawrence Erlbaum Associates.
Tunmer, W. E. & Nesdale, A. R. (1985). Phonemic segmentation skill and begining reading,
Journal of Educational Psychology 77: 417-427.
Valtin, R. (1984). Awareness of features and functions of language. In: J. Downing & R. Valtin
(eds.), Language Awareness and Learning to Read (pp. 227-260). New York: Springer
Verlag.
Yopp, H. K. (1988). The validity and reliability of phonemic awareness test, Reading Research
Quarterly 23: 159-177.

Address for correspondence: Marisol Carrillo, Facultad de Educaci6n, Campus de Espinardo,


E-30100 Murcia, Spain
Phone: +34 68 364 075; Fax: +34 68 364 146; E-mail: MsCarri@fc.um.es