Professional Documents
Culture Documents
Lundberg EffectsExtensiveProgram 1988
Lundberg EffectsExtensiveProgram 1988
Children
Author(s): Ingvar Lundberg, Jørgen Frost and Ole-Peter Petersen
Source: Reading Research Quarterly , Summer, 1988, Vol. 23, No. 3 (Summer, 1988), pp.
263-284
Published by: International Literacy Association and Wiley
REFERENCES
Linked references are available on JSTOR for this article:
https://www.jstor.org/stable/748042?seq=1&cid=pdf-
reference#references_tab_contents
You may need to log in to JSTOR to access the linked references.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms
Wiley and International Literacy Association are collaborating with JSTOR to digitize, preserve
and extend access to Reading Research Quarterly
University of Umeao
DANS LE BUT d'amener des enfants de niveau pr6scolaire a d6couvrir et a faire attention a la
structure phonologique des mots, les auteurs ont 6labor6 un programme d'enseignement
constitu6 d'exercices et de jeux m6talinguistiques. Pour 6valuer le programme ils ont men6
une 6tude longitudinale chez 235 enfants Danois de niveau pr6scolaire n'ayant requ aucun
enseignement de la lecture. Les enfants suivirent le programme d'entrafnement quotidien
pendant 8 mois. Un groupe contr61le de 155 enfants a 6galement subi des evaluations pr6-test
263
It has been demonstrated that instructional before the children had any formal reading in-
programs that incorporate phonemic trainingstruction, we have attempted to establish the
are more effective (e.g., Wallach & Wallach, causality in a less confounded setting.
1976; Williams, 1980). However, because such The limited focus on tasks involving rhyme
training has been an integral part of reading in- and alliteration in the Bradley and Bryant study
struction, it has been difficult to assess the puremay also have constrained its potential for gen-
effect of phonological awareness outside the eralization. The long-term effects on reading
context of formal reading instruction. It is only were also rather small: from a policy point of
with the landmark study of Bradley and Bryant view, perhaps too small to justify changes in
(1983) that the issue has been clarified. They current preschool curricula. The Olofsson and
studied the relationship between phonological Lundberg studies (1983, 1985) evaluated a con-
skills and reading acquisition with a combinedsiderably broader repertoire of phonological
longitudinal and experimental training designskills. However, because the training period was
that permitted a clear interpretation of the rather short, and because the training groups in-
causal direction. The longitudinal part involvedcluded only a few intact preschool classes, of-
some 400 children, who were followed over a fering a serious risk of confounding with
period of 4 years. The training study includedteacher factors, there were limitations to the
65 children who were 6 years old when a 2-year generalizability of the results.
period of phonological training started. One In the present article, we report a large lon-
subgroup of these children was trained in tasks gitudinal study in which some 400 children
in which they were to identify the odd one out were followed from kindergarten to Grade 2.
(e.g., in the four items bun, hut, gun, sun). ItOur intention was to overcome some of the limi-
could be demonstrated that early ability to han- tations of the earlier studies. The amount of
dle such tasks strongly predicted later reading training was more extensive than in earlier stud-
achievement, and that the training had a positiveies, and the scope of the metalinguistic training
effect on reading acquisition. Another study program was broader, comprising a gradual se-
(Olofsson & Lundberg, 1983) demonstratedries of exercises eventually leading to segmenta-
that it was possible to increase phonological tion and synthesis of phonemes. To evaluate the
awareness among preschool children by a rela-specific effects of the metalinguistic training,
tively short training period, of 6-8 weeks. Con-we also included a wider range of measures
tent, Morais, Alegria, and Bertelson (1982)than was used in earlier studies. Another feature
even found training effects after only four ses-
of the present study was our explicit effort to
avoid confounding with reading ability and
sions of training. And a later study by Olofsson
and Lundberg (1985) reported some-although reading instruction. A general problem con-
subtle-effects of such preschool training on nected with studies on phonological awareness
subsequent reading and spelling performance in
among preschool children has been the lack of
school. Although recent, sound research de-control for possible influences of early reading
signs such as those of Bradley and Bryant, Con- ability.
tent and associates, and Olofsson and Lundberg In Denmark, where this study was done,
have yielded important results, there is still children do not start school until the age of 7,
room for increased clarity. which means that the nonreading preschool
For example, one limitation of Bradley andchild has reached a more advanced stage of gen-
Bryant's training study was that phonemiceral cognitive development here than in educa-
awareness was taught while the children weretional systems where children start school
learning to read. In such a context, it is a bit earlier. Moreover, according to a long tradition
difficult to interpret the effectiveness of thein Denmark, preschool children are seldom
training program, because it interacts in an un-subjected to informal literacy socialization by
known way with the very teaching of reading.parents or older peers. Only in rare and excep-
By providing phonemic awareness training tional cases are childen able to read at the kin-
2. What is actually learned during the periment. The present solution was considered
metalinguistic training? to be reasonable, given the constraints imposed
3. Is the training effect lasting, and does by reality.
it transfer to new metalinguistic
tasks? Design
4. Does the preschool training facilitate In the beginning of the preschool year, all
reading and spelling acquisition in children in both groups were pretested with a
school? number of linguistic and metalinguistic tasks.
5. How specific is the training effect?Over the rest of the school year, from early Sep-
Does it affect aspects of general lan-tember to the end of May, children in the exper-
guage competence, or only skills in- imental group were given a training program
volved in phonemic analysis and comprising daily sessions of 15-20 minutes of
synthesis? metalinguistic exercises and games. The control
group followed the regular preschool program,
which in Denmark emphasizes social and aes-
thetic aspects of development and rather delib-
Experimental Control
M SD n M SD n F p
PRESCHOOL TESTS
PRETESTS
Prereading ability (4) 0.02 0.26 235 0.03 0.16 154 > I n.s.
Letter knowledge (28) 3.70 5.34 235 4.39 5.49 154 1.51 n.s.
Language comprehension (5) 3.14 1.02 230 3.12 0.99 154 > 1 n.s.
Vocabulary (66) 42.4 4.11 230 44.2 3.89 43 7.04 .01
Metaphonological tests
Rhyme (21) 15.8 3.93 235 16.1 3.93 155 > 1 n.s.
Word segmentation (2) 0.65 0.52 235 0.60 0.66 155 > 1 n.s.
Syllable synthesis (3) 2.06 1.04 235 1.83 1.10 155 4.53 .05
Syllable segmentation (3) 1.66 1.16 235 1.61 1.18 155 > 1 n.s.
Initial phoneme (8) 0.35 1.14 235 1.15 1.92 155 27.1 .001
Phoneme segmentation (8) 0.49 0.88 235 1.46 1.73 155 52.6 .001
Phoneme synthesis (8) 0.16 0.66 235 0.86 1.56 155 37.4 .001
Total combined
metaphonological (53) 21.1 6.15 235 23.6 7.46 155 12.5 .001
POSTTESTS
Prereading ability (4) 0.14 0.59 228 0.05 0.37 159 2.64 n.s.
Letter knowledge (28) 8.59 7.52 224 9.62 7.80 139 1.55 n.s.
Language comprehension (5) 3.58 0.91 220 3.51 0.85 156 > 1 n.s.
Vocabulary (66) 45.4 3.01 224 - - - - -
Metaphonological tests
Rhyme (21) 19.1 1.66 224 18.3 2.22 159 17.1 .001
Word segmentation (2) 1.10 0.43 224 0.77 0.54 159 44.4 .001
Syllable synthesis (3) 2.79 0.54 224 2.41 0.84 159 28.6 .001
Syllable segmentation (3) 2.57 0.81 224 2.26 0.90 159 11.9 .001
Initial phoneme (8) 4.88 2.96 225 1.81 2.58 159 111.6 .001
Phoneme segmentation (8) 3.21 2.27 224 1.73 1.94 159 44.5 .001
Phoneme synthesis (8) 1.84 2.18 224 1.05 1.74 153 13.5 .001
Total combined
metaphonological (53) 35.5 7.30 224 28.1 7.55 159 94.9 .001
SCHOOL TESTS
GRADE 1
Metaphonological tests
Rhyme (5) 4.96 0.24 210 4.55 0.94 103 33.7 .001
Syllable segmentation (6) 5.75 0.76 210 5.30 1.06 103 11.4 .001
Initial phoneme (5) 4.07 1.35 210 2.42 1.97 103 80.7 .001
Phoneme segmentation (9) 3.67 1.35 210 2.22 1.75 103 38.1 .001
Word length (5) 2.93 1.40 210 2.34 1.26 103 14.5 .001
Total combined
metaphonological (30) 21.4 3.69 210 16.8 4.27 103 84.5 .001
Raven's (21) 9.93 4.16 200 11.2 4.29 55 4.0 (.10)
Mathematics (48) 40.6 7.50 187 42.3 5.66 148 5.2 .05
Reading (400) 55.7 44.3 211 47.9 38.1 149 3.2 ( .10)
Spelling (28) 10.6 7.16 187 6.73 5.78 152 27.5 .001
GRADE 2
Reading (400) 124.7 74.9 185 104.4 71.1 135 6.2 .01
Spelling (28) 15.5 8.21 187 11.7 8.15 139 17.2 .001
Note. Maximum score is given in parentheses after each task.
Figure 1
Pretest to posttest scores on letter knowledge
for both groups
25 - EXPER I MENT
-5-- CONTROL
i20-
PRE POST
PRE POST
Figure 2
Pretest to posttest scores on language compre-
hension task for both groups
c3--- EXPERIMENT
- --G CONTROL
< 3-
0 !
PRE POST
Figure 3
Pretest to posttest total scores for all meta-
phonological tasks combined for both groups
0 -
13 - EXPERIMENT
5 - - CONTROL
25 -
z 20-
15
PRE POST
PRE POST
Figure 4a
Individual results for total combined meta-
phonological scores for experimental group
(Posttest results, on the y-axis, are plotted against pretest results, on
the x-axis. Points above the diagonal indicate positive change;
points on the diagonal, no change.)
EXPERIMENT
56
1 1
1 1 1 1
48 1 2 1 121 1
3 1 1 11
23 21 1 1
P 3 11 1 1
0 1 1 13 11 2 22 1
S 40 1 1 1111 1 1
T 1 1 1 111 21 11 1 1
T 1 1 1 232 11 4331 342
E 1 11 1 11 2 31 24
S 2 21 21 5 3 11 1 1
T 32 1 21 1 11
2 1 8 1314 1215
S 1 11 1 221 1 2 1
U 1 11 1
M 1 111 3 3 1 1 1
M 24
A 1 11
R 1 1
Y 1 11
16
0 4 8 12 16 20 24 28 32 36 40 44 48 52 56
PRETEST SUMMARY
testing with LISREL. The rhyme task was not to justify a subdivision of the metalinguistic
included in this analysis, because we had onlytasks when we further evaluate the treatment ef-
one subtest concerned with rhyming. With the fects in order to locate the most trainable di-
mension.
remaining 6 subtests, we obtained a nice confir-
mation of the model, as can be seen in Figure 5. Figures 6, 7, and 8 present the results with
The fit of the model is extremely strong, as indi-
this subdivision of tasks. As can be seen, by far
cated by the chi-square value, the adjusted the most dramatic effect is located at the pho-
goodness-of-fit index, and the root mean square
neme level (using a combined score from the
residual. Thus, the result of the analysis seems
three tasks involving phonemes). There was a
Figure 4b
Individual results for total combined
metaphonological scores for control group
(Posttest results, on the y-axis, are plotted against pretest results, on
the x-axis. Points above the diagonal indicate positive change;
points on the diagonal, no change.)
56
48
P 11
0 1 1
S 1 1
T 40 11
T 1 111
E 12 1 112 1 1
S 11
T 1 1 1 11 1112 211
32 1 1 1 1
S 1 1 1 1 2221
U 312 2 21 1
M 1 1 1
M 1 1 12 2221 413 1 2
A 24 1 1 2 1 1
R 2 1 111 11 I11 1 11
Y 3112 132 21
1
2 11
16 1 1
1
0 1
0 4 8 12 16 20 24 28 32 36 40 44 48 52 56
PRETEST SUMMARY
large Group x Test interaction effect, F(1, 359) We can now conclude that metaphonologi-
= 215.4, p < .0001. Although the correspond- cal training over an extensive period of time at
ing effects on word and syllable manipulation
preschool level has a positive but selective ef-
(using a combined score for the three word fect.
and It does not seem to promote general lan-
guage comprehension. It does not seem to
syllable tasks) and on rhyme recognition
(rhyme task) are more modest, there was still a the tendency to learn letters informally. It
affect
significant Group x Test interaction effect,
does affect metaphonological skills, in particu-
F(1, 373) = 7.22 and F(1, 373) = 7.82, re- lar those skills requiring the manipulation of
spectively; p < .01 for both. phonemes.
Figure 5
Factor analysis (LISREL) of model of
metaphonological abilities
WORD-
SEGMENTAT I ON
WORD
SYLLABLE
ANALYSIS
.z0
INITIAL
PHONEME
PHONEME . 59 PHONEME
SYNTHESIS AWARENESS
Figure 6
Pretest to posttest scores for rhyme test
for both groups
20.0-
13--- EXPERIMENT
- CONTROL
15.0
z 10.0
cn
S.O-
0.0-
PRE POST
Figure 7
Pretest to posttest combined scores for three
word and syllable segmentation tasks
for both groups
.0 -
E-3 - EXPER I MEN T
6.0 - - CONTROL
z
" S.O-
0 Z.0-
CD 3.0
> 2.0
1.0
0.0-
PRE POST
Figure 8
Pretest to posttest combined scores for three
phonemic segmentation tasks for both groups
10.0 -
9.0- -E EXPERIMENT
- CONTROL
8.0-
z
- 7.0-
o 6.0-
S.O-
UL
> 3.0
2.0-
1.0
0.0-
PRE POST
group inoutperformed
trol group actually mathematics, F(1, 333) = 5.2, th
mental group. However, the
p < .05. The direction of small
this difference sug- nu
control children who
gests that took
the preschool the
training effect was spe- Ra
should be noted. cific to literacy tasks rather than having a
Because one reason for including the Rav- general impact on school-related activities.
en's in the assessment program was to check for
the specificity of the effects, the extent to whichEffects of training on reading and spelling
performance on the Raven's correlates within school
other variables should be informative. In most
cases, the product-moment correlations were As indicated in Table 1, there were signifi-
low and insignificant, with the following excep-cant differences between the experimental
tions: language comprehension (.30 at pretest, group and the control group in both reading and
spelling in Grades 1 and 2, although the differ-
.28 at posttest), vocabulary (. 19 at pretest, .22
ence in reading in Grade I was only marginally
at posttest), letter knowledge (. 15 at pretest, .22
at posttest), syllable synthesis (.23 at posttest),
significant (p < .10). In a 2 x 2 ANOVA of
and the total combined score for phonological reading scores with group and grade as varia-
awareness (.22 at posttest; p < .01 for all). The bles, there was, of course, a strong main effect
generally low correlations suggest that we areof grade, F(1, 318) = 497.9, p < .0001; and
dealing with a specific and relatively pure fac-also a significant effect of group (the experi-
mental group outperformed the control), F(1,
tor, something like a module for operations on
language, which has a weak relation to lan-318) = 5.00, p < .05. There was also a signifi-
guage use itself and to cognitive ability.
cant Group x Grade interaction, F(1, 318) =
5.75, p < .05, which here means that the effect
Mathematics test. Table 1 indicates that the
was even greater at Grade 2 (see Figure 9).
control group outperformed the experimental
Figure 9
Mean number of words correctly read in
Grade 1 and Grade 2 for both groups
WORDS
120
100 -
80
60 -
40 -
20
GRADE 1 GRADE 2
Figure 10
Mean number of words correctly spelled in
Grade 1 and Grade 2 for both groups
WORDS
20 -0 EHPERIMENT
te18 CONTROL
16
14
12-
10-
GRROADE 1 GRROADE 2
Yun-Fei, Hong-Yin,
The slightly higher performance noted on and Bao-Qing (1986) have
the math test for the control children from
demonstrated the Jut-
crucial role of an alphabetic
land was in agreement with earlier script asnational
the basis foras-
developing segmentation
sessments, where Jutland tends skill.
to In outperform
this study, we seem to have found con-
Bornholm on the same reading tradictory
test as evidence:
the one Phonological skill can be
used here. Apparently, the preschool developed outside the context of formal reading
training
for the Bornholm children caused a clear shift instruction. But the contradiction is probably
in this pattern of traditional inferiority as far asmore apparent than real. The crucial factor
literacy was concerned. seems to be explicit instruction, rather than spe-
In the case of spelling, the effect was evencific encounters with the letters of the alphabet.
clearer. The overall difference between the two The design of the present experiment per-
groups was significant, F(1, 301) = 20.9, mitted us to localize the training effect specifi-
p < .0001. There was again, of course, a sig- cally to phonological awareness. Functional
nificant effect of grade, F(1, 301) = 1.3, p < linguistic competence, such as the ability to fol-
.001, but there was no significant Group x low verbal instructions, was not affected by the
Grade interaction effect, F(1, 301) = 0.06, p training program. Neither was the acquisition of
> .10 (see Figure 10). letter knowledge. Within the metalinguistic do-
Another way of demonstrating the impact main, we also observed differential effects. Al-
of metaphonological ability in preschool on though statistically significant, the effects on
later reading and spelling performance is the rhyming and on word and syllable awareness
use of multiple regression analysis. With read- were comparatively modest, whereas the effect
ing performance in Grade 2 as the criterion, on ability to perform phonemic tasks was quite
only two independent variables from the pre- dramatic.
school assessments entered the equation. At the The development of rhyming ability does
first step, the combined performance on the not seem to be strongly dependent on formal
three phonemic tasks at posttest yielded an R of training. Cary, Morais, and Bertelson (1987)
.58. The second step included language com- have observed rhyming and alliteration in an il-
prehension at posttest, and the R increased to literate poet. Rhyming seems to require less
.60. For spelling in Grade 2, only phonemic conscious and deliberate manipulation of seg-
skills entered the equation, for an R of .61. All ments. The attention is more directed to global
remaining independent variables were insignifi- similarity. Morais et al. (1986) suggest the term
cant. Thus, the level of phonemic ability in pre- sensitivity to sound similarity to describe the
school is a powerful predictor of reading and ability involved in solving rhyming tasks.
spelling performance in school. Awareness of words and syllables is evi-
dently also less dependent on systematic in-
struction for its development. As compared to
phonemes, syllables are more accessible units
Discussion
of the speech signal-more isolable, more sali-
ent, and less abstract. In order to attend to sylla-
bles, the child does not have to ignore the
The evidence obtained in the present study
suggests, first, that phonological awarenessnatural
can unity of the articulatory act. A mini-
be developed before reading ability and inde-
mum of guidance, then, should be required for
the child to manipulate them (see Mann, 1986).
pendently of it, and, second, that this phonolog-
ical awareness facilitates subsequent reading Phonemic segments, however, do not seem
acquisition, thus providing unconfounded to be spontaneously available for conscious at-
evi-
dence of a causal link. tention, as was reflected here in the low per-
However, segmentation ability does not formance on the phonemic tasks at pretest by
both groups, and at posttest by the control
seem to develop spontaneously. Morais, Ber-
telson, Cary, and Alegria (1986) and Read,group. The experimental group evidently bene-
STANOVICH, K.E., CUNNINGHAM, A., & CRAMER, B. (1984). analysis and phoneme blendin
sis on phoneme
Assessing phonological awareness in kindergarten chil-
nal of Educational Psychology, 72, 1-15.
dren: Issues of task comparability. Journal of Experi-
Footnotes
mental Child Psycholog)y 38, 175-190.
TORNtUS, M. (1984). Phonological awareness Weand
gratefully
reading:acknowledge the invaluable help from
A chicken and egg problem? Journalthe ofteachers
Educational
and consultantsoin Bornholm and Jutland who
Psychology,, 76, 1346-1358. participated in this project. Ake Olofsson of the University
TUNMER, W.E. (1986, March). Cognitive andof UmeA assistedfac-
linguistic with the statistical work. Peter Bryant
gave
tors in learning to read. Paper presented atvaluable suggestions in a review of an earlier version
the Confer-
of the manuscript.
ence on Early Reading, Center for Cognitive Science, The project had financial support from
University of Texas at Austin. the local school board of Bornholm and from the Swedish
National Board
WALLACH, M., & WALLACH, L. (1976). Teaching of Education.
all children
to read. Chicago: University of Chicago Press.
VALTIN, R. (1984). The development of metalinguistic abili-
ties in children learning to read and write. In J. Downing
& R. Valtin (Eds.), Language awareness and learning to Received August 18, 1987
read (pp. 207-226). Berlin: Springer-Verlag. Revision received January 12, 1988
WILLIAMS, J.P. (1980). Teaching decoding with an empha-Accepted February 23, 1988