See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/20396377

Nursery Rhymes, Phonological Skills and Reading

Article  in  Journal of Child Language · July 1989

DOI: 10.1017/S0305000900010485 · Source: PubMed

CITATIONS READS

256 5,630

4 authors, including:

Morag Maclean
Oxford Brookes University
28 PUBLICATIONS   1,977 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Parental reflective function and children's emotion regulation View project

All content following this page was uploaded by Morag Maclean on 18 June 2014.

The user has requested enhancement of the downloaded file.

Journal of Child Language
http://journals.cambridge.org/JCL

Additional services for Journal of Child

Language:

Email alerts: Click here

Subscriptions: Click here
Commercial reprints: Click here
Terms of use : Click here

Nursery rhymes, phonological skills and reading

P. E. Bryant, L. Bradley, M. Maclean and J. Crossland

Journal of Child Language / Volume 16 / Issue 02 / June 1989, pp 407 - 428

DOI: 10.1017/S0305000900010485, Published online: 17 February 2009

Link to this article: http://journals.cambridge.org/abstract_S0305000900010485

How to cite this article:

P. E. Bryant, L. Bradley, M. Maclean and J. Crossland (1989). Nursery rhymes,
phonological skills and reading. Journal of Child Language, 16, pp 407-428
doi:10.1017/S0305000900010485

Request Permissions : Click here

Downloaded from http://journals.cambridge.org/JCL, IP address: 161.73.106.138 on 18 Jun 2014

J. Child Lang. 16 (1989), 407-428. Printed in Great Britain

Nursery rhymes, phonological skills and reading*

P. E. BRYANT, L. BRADLEY, M. MACLEAN
AND J. CROSSLAND

University of Oxford

(Received 18 February 1988. Revised 9 July 1988)

ABSTRACT
Nursery rhymes are an almost universal part of young English-speaking
children's lives. We have already established that there are strong links
between children's early knowledge of nursery rhymes at 3;3 and their
developing phonological skills over the next year and a quarter. Since
such skills are known to be related to children's success in learning to
read, this result suggests the hypothesis that acquaintance with nursery
rhymes might also affect children's reading. We now report longitudinal
data from a group of 64 children from the age of 314 to 6;3 which
support this hypothesis. There is a strong relation between early
knowledge of nursery rhymes and success in reading and spelling over
the next three years even after differences in social background, I.Q and
the children's phonological skills at the start of the project are taken into
account. This raises the question of how nursery rhymes have such an
effect. Our answer is that knowledge of nursery rhymes enhances
children's phonological sensitivity which in turn helps them to learn to
read. This paper presents further analyses which support the idea of this
path from nursery rhymes to reading. Nursery rhymes are related to the
child's subsequent sensitivity to rhyme and phonemes. Moreover the
connection between knowledge of nursery rhymes and reading and
spelling ability disappears when controls are made for differences in
these subsequent phonological skills.

[*] This research is supported by a research Grant from the Medical Research Council. We
are also grateful for the help of Marie Clay, Chris Pratt and Usha Goswami, all of whom
read and commented on an earlier version of this paper. We should like to thank Kathryn
Rhodes for her part in helping to collect data during the first year of our project. We are
very grateful to the teachers and staff of several local Primary and First schools for letting
us visit the children in our project at school. The schools are: Bernwood, Botley, Church
Cowley, East Oxford, Greycoates, Headington Quarry, Kennington, Larkrise, New
Marston, Orchard Meadow, Our Lady's, Pegasus, St Andrews, St Ebbes, St Francis, St
John Fisher, St Josephs, St Mary and St John, St Nicholas, St Thomas More, Sandhills,
Speedwell, Speenhamland, The Crescent, The Queen's Dyke, Windmill, Wolvercote and
Wood Farm. Address for correspondence: P. E. Bryant, Department of Experimental
Psychology, University of Oxford, South Parks Road, Oxford OXi 3UD, UK.

15 407 JCL 16
 CHILD LANGUAGE

INTRODUCTION
Interest has been growing in the role of linguistic routines such as nursery
rhymes, action rhymes and word games in children's development. They are
pervasive, they obviously play a significant part in family life and particularly
in interactions between parents and their young children (Snow, De Blauw
& Van Roosmalen 1979), and they have stood the test of time remarkably
well. Recent work by Trevarthen (1986, 1987) has shown that mothers do
recite nursery rhymes to infants as young as three months, and has also
established that there are striking temporal regularities in the way in which
mothers sing or speak nursery rhymes and lullabies. It has also been found
that children, including children with Down's syndrome, enjoy nursery
rhymes and deliberately choose to listen to them from a very early age
(Glenn, Cunningham & Joyce 1981). Yet, as far as we know, no psychologist
has investigated the possibility that young children's knowledge of nursery
rhymes could affect their linguistic development. Chukovsky (1963) did
suggest that poems of various kinds are an important part of the young child's
experiences, but no one has taken up this idea in systematic research.
What kind of influence could nursery rhymes have ? They may help
syntactic or semantic development. However, the language in them is often
as simple, syntactically and semantically, as anything they would hear in
other contexts (Humpty Dumpty sat on a wall; the mouse ran up the clock),
or incomprehensible to the parent as well as to the child (Hickory dickory
dock).
There is another possibility, and that is that they enhance children's
phonological skills. Rhymes (Jack and Jill went up the hill) and half rhymes
(Goosey goosey gander where do you wander) are of course a necessary,
frequent and usually heavily stressed part of nursery rhymes. So it is quite
possible that young children could learn about rhyme with the help of these
traditional routines. Rhyme and alliteration are significant phonological
phenomena. A child who is sensitive to rhyme and alliteration must recognize
at some level (though not necessarily explicitly) that different words and
different syllables have a segment of sound in common. Cat and hat, for
example, rhyme because they share the common sound at.
Children have this sensitivity long before they go to school. Four-year-
(Knafle 1973, 1974, Lenel & Cantor 1981) and even three-year-old children
(Maclean, Bryant & Bradley 1987) perform well above chance level in rhyme
detection tasks. In fact rhyme and alliteration tests are the only measures of
phonological sensitivity which one can rely on to produce above-chance level
results in preschool children. Tasks which involve detection of single
phonemes are usually too difficult for children who have not yet learned to
read (Bruce 1964, Liberman, Shankweiler, Fischer & Carter 1974, Liberman,
Shankweiler, Liberman, Fowler & Fischer 1978, Bryant & Goswami 1987).
408
 NURSERY RHYMES AND READING

They also cause a great deal of difficulty to adults, and to older children, who
have not learned an alphabetic script (Morais, Cary, Alegria & Bertelson
1979, Mann 1986, Morais, Bertelson, Cary & Alegria 1986, Read, Zhang, Nie
& Ding 1986).
The possibility that knowledge of nursery rhymes affects children's
sensitivity to rhyme and alliteration suggests another connection. Nursery
rhymes may also play a part when children learn to read. There is
considerable evidence that the sensitivity to rhyme and alliteration which
children acquire before they go to school does play a causal role in their
reading several years later on (Bradley & Bryant 1983, 1985). It follows that
if nursery rhymes do make the young child more sensitive to rhyme and
alliteration they should in the end have a similar effect on that child's
progress in reading.
Some evidence on these questions has already been provided in a
longitudinal report (Maclean et al. 1987) which covered a 15-month period
and began when the children were 3:4. At this age we measured the
children's knowledge of five extremely common nursery rhymes. Nearly all
the children knew something of some of these nursery rhymes, and the extent
of their knowledge then was a powerful predictor of their growing skill in
rhyme and alliteration detection tasks over the next year and a quarter. This
was the first systematic evidence that knowledge of nursery rhymes might
play a role in children's phonological development. However it covered a
relatively brief period and it did not establish a definite link with reading,
because at the end of the period reported the children were only 4; 7 and most
of them could not read anything at all.
This paper now reports longitudinal data from the same study over a
three-year period, from the beginning of the project when the children were
3; 4 to a time when as six-year-olds most of them had been in school for a year
and had begun to learn to read. Our aim was to establish the relationships
between the children's original knowledge of nursery rhymes and their
progress several years later in learning to read and to spell. We also wanted
to know whether any link between these two skills can be explained in terms
of the effects of knowledge of nursery rhymes on children's phonological
sensitivity.

METHOD
Subjects
Ages. There were 66 children in this project, but data are reported on 64 (31
boys and 33 girls). One child was left out because we failed to test her
knowledge of nursery rhymes when the project began; the other child left the
country half way through the project. All but one of the children came from
native English-speaking backgrounds. The exception was a boy whose
409 is-2
 CHILD LANGUAGE

TABLE i. Social class and parental education in our group

SOCIAL CLASS
National Group' s
Occupations National %1 Group's % N
Professional 56 3i 2
2. Intermediate 18-4 359 23
3 N. Non-manual skilled 21-5 109 7
3M . Manual skilled 3'i 328 21
4- Manual partly skilled 177 94 6
5- Unskilled 57 0 0
6. Single-parent families 79 5
Total 64
MOTHERS' EDUCATIONAL LEVEL

Qualification" % N

1. Degree(s) 188 12
2. HNC/Cert.Ed. 94 6
3- A/ONC io*9 7
4- O/CSE 359 23
5- None 250 16
Total 64

HNC = Higher National Certificate, A = GCE Advanced level, ONC = Ordinary

National Certificate, O = GCE Ordinary level, CSE = Certificate of Secondary Education.

mother is Swedish: though English is the language spoken in his home he

knows a certain amount of Swedish as well. At the start the average age of the
64 children was 3;4 (range 2; 10—3;9). This study spans three years so when
the last measure was taken the average age of the 64 children was 6; 3 (range
5; 9-6; 8).

Social Background. The children came from a wide range of backgrounds.

Measures of the home background included social class and the educational
level of the parents, and questionnaires given to the parents at the end of the
first and second years of the project on the amount of linguistic support for
the child at home. Table 1 gives the social class and also the educational level
of the parents.
In the first questionnaire the parents were asked how many books the child
owned. In the second they were asked the same question, and also to indicate
on a seven point rating scale 'Whose responsibility is it to teach a child... ?'
various skills. These included 'letter names', 'to read' and 'to write'. The
points on the rating scale were that the responsibility lay with: (1) parent; (2)
mainly parent; (3) more parent than school; (4) parent and school equally;
(5) more school than parent; (6) mainly school; (7) school.
410
 NURSERY RHYMES AND READING

We decided to use mothers' educational level as our main measure of the

children's background. We could not use social class because the project
included several single parent (mother) families to whom it was not possible
to apply the social class index because it is based on the father's occupation.
The correlation matrix in Table 2 shows the relationships between our
reading measure at 6; 3, maternal education (treated as a continuous variable)
and these questionnaire measures. In all cases the maternal education
measure correlates more highly with the children's final reading levels than
do the questionnaire variables. Thus mothers' educational level was the most
powerful index of the possible effects of the children's different backgrounds.

Procedure
The project is a longitudinal one and our measures are predictive. We shall
report the relationship between four sets of measures taken over five sessions.
The first two of these sessions were in the children's homes, but in the last
three sessions the children were seen in their schools. Our aim was to
discover the relationship between the following four sets of measures:

(1) The children's knowledge of nursery rhymes

This was measured at the start of the project when the average age of the
children was 3; 4. We selected five of the most popular nursery rhymes and
asked each child if s/he could recite them. We had established the relative
popularity of these nursery rhymes in a pilot study. The five were Humpty
Dumpty, Baa-Baa Black Sheep, Hickory Dickory Dock, Jack and Jill and
Twinkle Twinkle Little Star. Our request to the child, with each of the five
nursery rhymes, was: 'Can you say (e.g.) Humpty-Dumpty for me?' Then
we recorded whether the child knew the whole rhyme, or part of it or none
of it at all.

(2) Phonological sensitivity

Rhyme detection. In the first two years the children were given rhyme
detection tests in which the rhyming words ended with the same vowel-
consonant combination (e.g. peg leg). We shall report the results of two of
these tests, one given when the children were 3 ;4 and the other given when
they were 4; 7.
We devised a version of the rhyme-oddity task that we have used in
previous studies (Bradley 1980, Bradley & Bryant 1983). The new thing
about this version was that we used pictures in order to remove the memory
load from the task. Before the trials began we asked the child if they knew the
nursery rhyme 'Jack and Jill', and recited the first two lines. 'Jack and Jill
411
 T A B L E 2. Home background measures and children's reading level at 6 ; J : correlation matrix

Parents should teach child

Children's
child's child's 1 2 3 reading
books at 4 yrs books at 5 yrs letters to read to write level at 6; 3
Mothers' 0-441*** 0-471*** 0288* 0-260* 0183 o-6o7**»
educational (6o) (63)
level
Number of 0-538*** 0-327** oigi 0160 O-4JI***
child's (59) (60) (60) (60) (60)
books at 4 yrs
Number of 0-244* 0058 0039 0-484***
child's (63) (63) (63) (63)
books at 5 yrs
Parents 0-45O*** 0-232* 0-322**
should teach
C H I L D L A N G lI A G E

child letters
Parents 0-703*** 01 go
should teach
child to read
Parents 0008
should teach
child to write

*, p < 0-05; •*, p < o-oi; ••*, p < o-ooi.

Numbers in brackets indicate N for measure if less than 64.
 NURSERY RHYMES AND READING

went up the...yes 'hill'. Jill, hill, they sound the same, they rhyme, can you
tell me another word that sounds like hill ? ... fill. Now we're going to play a
game about words that sound the same, about words that rhyme.' On each
occasion the test consisted of two practice trials and then ten experimental
trials. In each trial the child was given three words, with pictures, two of
which rhymed while the third did not (e.g. peg cot leg; fish dish book). The
child's task was to tell us the one that did not rhyme.

Phoneme oddity. In the third and fourth years we adapted the same oddity
method to deal with the detection of single phonemes. The three main
phoneme oddity scores that are reported here come from two tests given
when the children were 5;7 and a third when they were 6;3. At 5;7 we gave
them an OPENING PHONEME test in which they had to recognize which words
began with the same phoneme (e.g. peg pot) and an END PHONEME test in
which they had to recognize words which ended with the same phoneme (e.g.
pin gun).

Opening phoneme test (517). The children were asked to repeat a set of four
words, then to say which word did not sound like the others. The odd word
out began with a different sound from the other words (e.g. peg land pin pot;
leg bus band boat). Corrective feedback was given on the first two trials,
followed by five further trials.

End phoneme test (5; 7). The same procedure was followed, but in each trial
the odd word ended with a different sound from the other words (e.g. pin gun
hat men; met ball sell doll).

Opening phoneme test (6; 3). At the beginning of the fourth year when the
children were 6;3 they were given another phoneme oddity test. This time
they were given corrective feedback on three practice trials before attempting
ten test trials (e.g. peg pain side; page dog pen).

Object naming (4; 11). This was a non-phonological test. For control purposes
we needed a non-phonological test which is related to reading ability. We
chose object naming because there is strong evidence that it distinguishes
good from poor readers (Denckla & Rudel 1976). The children were shown
a set of ten unambiguous pictures, cup, table, tree, boat, knife, key, window,
finger, duck and snake. They were asked to name the things on the board as
fast as they could without making mistakes. They were timed from their first
response to their last response.

413
 CHILD LANGUAGE

(3) Reading and spelling

Reading. We used the Spar Reading test which has two forms. The children
were given Spar Form A when they were 5; 11, and Spar Form B when they
were 6; 3. The test is divided into two parts. The first section has 15 pictures
with four or five words beside each picture (e.g. picture of a car with the
words: cake camp car came). The children are shown how to underline the
word that goes with the picture on three practice items, and then complete
the test items without help. If they complete this section without making ten
errors they are shown how to do three practice items for the second part of
the test. This time they have to underline the correct one out of six words to
complete a short sentence correctly (e.g. Lemons are year yes you yet yellow
yard). There are 30 test sentences. Testing stops when the child has made ten
errors from the beginning of the test. One point is given for each correct
answer; we used these raw scores in our analyses.

Spelling. The children were given the Spar Spelling test when they were
5 ; 11. We said each test word once, repeated the word in a sentence, and then
said the word again before the child wrote the word. We stopped testing
when the child had made six consecutive errors.

(4) I.Q. and vocabulary (B.P.V.S.)

We gave the children the British Picture Vocabulary Scale (a version of the
Peabody Picture Vocabulary test standardized in Britain) at the beginning of
the project when they were 3; 4. The mean ratio score (average for the
population is 100) on the B.P.V.S. was 104 (s.D. 12-81). An I.Q. test (the
W.P.P.S.I.) was given at 4;3. The mean I.Q. was n r o 8 (s.D. 12-22).

RESULTS

We shall describe the children's level of performance in the various tasks

before dealing with the relationships between them.

Levels of performance
Knowledge of nursery rhymes. We used a composite score which combined the
children's partial and complete knowledge, with a score of 1 for each partial
production of a nursery rhyme and 2 for a complete version. Thus the
maximum possible score for the five trials was 10. The mean score was 4-44
(s.D. 2-36). Only one child scored o, and so very nearly all the children in our
group knew something about nursery rhymes at the very young age of 314.
The scores were normally distributed (see Maclean et al. 1987).
414
 NURSERY RHYMES AND READING

Detection of rhyme. At 354 the mean for our sample of 64 children was 4-84
out of 10 (s.D. 2-22). Since there were three choices chance level in this task
would be at 3-3 correct. An individual score of 7 or more would (according
to the binomial) be significantly above chance level (p < 0-05). Fourteen of
our 64 children reached this level. Thirty-five of the group of 64 scored 4 or
below. Thus there was a slight negative skew to these scores and we were not
confident that they could be treated as a continuous variable.
At 4;7 the scores were a great deal higher — mean 6#5O (s.D. 2^40) — and
normally distributed. Thirty-one of the children achieved a score of 7 or

Detection of phonemes. At s;7 the children's scores (out of 7) were 4-62 (s.D.
2-19) in the opening phoneme test and 2-94 (s.D. i-7o) in the end phoneme
test. Twenty of the children achieved the maximum score of 7 in the easier
opening phoneme test which gave the scores a positive skew. So this variable
was not treated as a continuous one. The scores in the end phoneme test were
normally distributed.
At 6;3 the children's scores (out of 10) in the opening phoneme test were
6-94 (s.D. 2"66) and were normally distributed.

Reading and spelling. At 5; 11 it was clear that many of the children in our
group could neither read nor write, and that meant a pronounced bimodal
distribution in our reading and spelling scores. Thirty-one of the children
could spell no words in the spelling test. Thirty of the rest spelled between
one and seven words correctly. We decided to treat this as a binary,
dichotomous variable. It was much the same pattern with the reading ages.
Thirty of the children showed no real sign of reading in this test. The
number of words that the children read in this test was also made into a
dichotomous variable.
At 6;3 the scores were much more normally distributed. The children's
mean reading age was 6; 8 (s.D. 11 '96). The mean number of words read was
1 1 57 (s.D. 9-17). The number of words that the children read was treated as
a continuous variable.

The relationship between early knowledge of nursery rhymes and success in

learning to read
Table 3 gives the correlations between the measures of the extent of
children's knowledge of nursery rhymes and their success in reading and
spelling as well as their scores in the tests of phonological sensitivity. These
were all strong correlations, and it is very interesting to see how well the
nursery rhyme score correlated with later phonological measures and with
the reading test given three years later (r = 0-56, p < o-ooi).
415
 CHILD LANGUAGE

TABLE 3. Correlations between continuous variables

Rhyme Phoneme Phoneme

detection detection detection Reading
(4; 7) (5; 7) (6; 3) (6; 3 ) I.Q. B.P.V.S.
Nursery o-64*« o-6i»" o.so«. O.S9... 0-30**
°'39***
rhyme
knowledge
(3; 4)
Rhyme 061***
detection
(4; 7)
Phoneme o-64### 058*" o-46 ###
detection
(Si 7)
Phoneme o-66*»» 061 ••• o-34 ##
detection
(6; 3)
Reading o-66«* O-42* # #
(6; 3 )
I.Q. 0-34**

*, p < 005; •*, p < 001; ••*, p < 0001.

However simple correlations on their own are not convincing evidence of

a strong connection between nursery rhymes and reading or phonological
skills. It is quite possible that the two correlated variables could both be
determined by differences in some other factor, such as I.Q. The children
who know more nursery rhymes at three and read well at six may do so
because they have a higher I.Q. In order to rule out the effects of extraneous
variables of this sort, we turned to fixed-order multiple regressions and to
logistic analyses.
In a fixed-order multiple regression there is a dependent variable (for
example, success in reading) and several independent variables. Assuming
that all but one of these independent variables represent extraneous factors
such as I.Q. and social background whose effects one wants to control, and
that the remaining independent variable is the one of interest, then one enters
the extraneous variables first in successive steps, entering the crucial
independent variable as the final step. By this stage, all the variance due to
the extraneous variables has been accounted for, and so if the final step does
account for a significant amount of the variance, one can be sure that there
is a genuine connection between it and the dependent variable.
Fixed-order multiple regressions are appropriate when the dependent
variable is normally distributed. Our last measure of reading, at 6; 3, did have
a normal distribution, but our earlier measures of reading and spelling at 5; 11
416
 NURSERY RHYMES AND READING

(as well as one of our measures of phoneme detection) did not. We

dichotomized the reading and spelling scores, putting those who read (or
spelled) something into one category and those who showed no sign of
reading or spelling into another. To analyse these dependent variables we
turned to logistic analyses, using the GLIM (Generalised Linear Interactive
Modelling) system. Apart from the difference in the nature of the dependent
variable, the procedure in these analyses is exactly the same as in the multiple
regressions. In logistic analyses the statistic which indicates whether each
independent variable has a significant effect is the / 2 , in comparison with
multiple regressions where it is F.
The independent variables were entered in the same order in all these
analyses. Each analysis contained six independent variables entered in six
successive steps. The first five were extraneous variables. The six steps,
which can be seen in the section headed ' Direct relations' in Table 4, were
as follows.
(1) The children's ages at the time of the reading or spelling test.
(2) Their I.Q.
(3) Their vocabulary (B.P.V.S.) at 354.
(4) The mother's educational level: this was used as the measure of the
differences in the children's backgrounds. There were five different edu-
cational levels, but we could not treat them as a linear variable, because it is
risky to assume, for example, that the difference between levels 1 and 2 is the
same as between levels 2 and 3.
(5) The children's sensitivity to rhyme at 314 was tested at the same time
as their knowledge of nursery rhymes. It is quite possible that knowledge of
nursery rhymes could simply be a measure of phonological sensitivity and
thus that any relation between nursery rhyme knowledge and reading simply
means that children who have a greater phonological sensitivity at three
eventually become better readers. We wanted to ensure that there was more
to any connection between nursery rhymes and reading than this and so we
included the original rhyme detection scores — our measure of phonological
sensitivity at the time — as our fourth extraneous variable. These scores were
not normally distributed: about half the children scored 3 (which is roughly
chance level) or less. We made the variable a dichotomous one, dividing
the children into those who scored 3 or below and those who scored more
than 3.
(6) The child's knowledge of nursery rhymes when he or she was 314.
This, our final step, was the variable that we predicted would have a
significant connection with reading and spelling two to three years later.
The first section of Table 4 (Direct relations) gives the results of these
analyses. It shows that the children's knowledge of nursery rhymes did
indeed predict their success in reading and spelling two to three years later
on, even after the effects of the three extraneous variables had been ruled out.
417
 CHILD LANGUAGE

TABLE 4. The relation between nursery rhyme knowledge and reading and
spelling (six- and seven-step fixed-order multiple regressions and logistic
analyses)

Reading-5 ; 11 Spelling-5; 11
Change Change Reading-6; 3
Dependent in scaled in scaled Change
variable deviation (/ 2 ) deviation (^2) infl 2 F

Direct relations: six steps

Step ia 8-Q7" 1 09 007 4-73*
Age at read/spell
test
Step 2 27-831 • • • 25-ISI*** °'4S 56-98*»«
I.Q. (W.P.P.S.I.)
Step 3 017 O-2I O-O2 328
Vocab (314)
(B.P.V.S.)
Step 4 b 1229* 331 0O97 379*
Mothers' educational
level
Step 5 029 406 O0O8 129
Rhyme detection
(3;4)
Step 6 4'47* 980" OO3 4-22*
Nursery rhyme test
(3; 4)
Controlling for rhyme: seven steps
Steps 1-5 As before
Step 6 io-77** 587* OOS 8-47"
Rhyme detection
(4; 7)
Step 7 009 3'74 0003 047
Nursery rhyme test
(3; 4)
*, p < 0 0 5 ; * * , / > < 0 0 1 ; * * * , p < 0 0 0 1 .
" degree of freedom for steps 1-3 and 5-7 is 1.
b
degrees of freedom are 4.

The connection cannot be the result of differences in the children's

intelligence or in their social background or even in their initial phonological
sensitivity, because these variables were controlled.
This result is a striking one, given that the controls are so stringent and the
distance in time between one measure (nursery rhymes) and the other
(reading and spelling) so great. It suggests that early, entirely informal,
experience with (or at least knowledge of) linguistic routines such as nursery
418
 NURSERY RHYMES AND READING

rhymes does play a considerable role in preparing children for learning how
to read and write.

Possible connections between nursery rhyme knowledge and reading ability

How do nursery rhymes prepare children for reading ? Our hypothesis is that
nursery rhymes enhance children's sensitivity to rhyme and alliteration and
thus increase their phonological sensitivity. The hypothesis leads to two
predictions.
The first is that the early nursery rhyme knowledge scores will also predict
the extent of the children's later phonological sensitivity even when their
initial phonological skills (at the time of the nursery rhyme measures) are
controlled.
The second and less obvious prediction is that the early nursery rhyme
scores will not be related to reading and spelling later on if one controls for
the effects of phonological sensitivity at the time of the later reading or
spelling measure. If, as we claim, nursery rhymes enhance reading via the
child's phonological sensitivity then controlling for the extent of phonological
sensitivity should remove the connection between nursery rhymes and
reading.

First prediction
There is already some support for the first prediction from this project. In an
earlier paper (Maclean et al. 1987) we showed that the nursery rhyme scores
do predict rhyme detection at 4;7 even when I.Q., social background,
vocabulary and the children's initial sensitivity to rhyme have been con-
trolled. In the current study we went on to test the same prediction further
by looking at the relation between the early nursery rhyme scores and the
three phoneme detection tasks.
Our analyses were either fixed-order multiple regressions or logistic
analyses. The dependent variables were the scores in the later phoneme
categorisation tests. There were three analyses and thus three dependent
variables: the opening phoneme and the end phoneme tests at 5:7 and the
opening phoneme test given at 653. The distribution of the scores in the
opening phoneme test at 5; 7 was skewed. The other two dependent variables
were normally distributed. So we applied a logistic analysis to the first of
these dependent variables and fixed-order multiple regressions to the other
two.
The independent variables were the same six as in the previous analyses,
and were entered in the same order as before. Thus the purpose of these
analyses was to establish whether the children's knowledge of nursery
rhymes predicted their phonological sensitivity two to three years later after
419
 CHILD LANGUAGE

controls for differences in their intelligence, social background and also in

their initial level of phonological sensitivity.
The results of these analyses are presented in Table 5 and show a
considerable and impressive connection between nursery rhyme knowledge
and the development of phonological sensitivity over the next two years. The
significant relation between nursery rhymes at three and the two tests of
phoneme detection at five years suggests that children's knowledge of
nursery rhymes does enhance their awareness of the component sounds in
words, even when the sound in question is a single phoneme. The relation fell
just short of significance in the case of the phoneme detection task at 6; 3. We
are not sure why this is. It may be that by this time other influences, and
particularly the experience of learning to read, are having a strong effect on
sensitivity to phonemes.
The significant relation between nursery rhymes and sensitivity to rhyme
(4; 7) and to phonemes (5; 7) allows us to ask whether the pathway between
nursery rhymes and reading is through phonological sensitivity as measured
by these tests. This question leads to our second prediction.

Second prediction
If the pathway is through phonological sensitivity, then the relationship
between nursery rhyme knowledge and reading ability should disappear if
one of these measures (rhyme at 4; 7 or phoneme detection at 5 ; 7) is included
as an independent variable and as a prior step in the multiple regression or
logistic analysis. We tested the prediction in two sets of analyses in which the
dependent variables were the same reading and spelling tests as before. The
first five steps in all these analyses were exactly as before. The sixth was one
of the phonological tests (rhyme at 417 or phoneme detection at s;7 and in
one analysis at 6; 3). The final step was the nursery rhyme scores. Thus the
seven steps were: (1) the children's age at the time of the reading or spelling
test; (2) I.Q.; (3) B.P.V.S.; (4) the mothers' educational level; (5) the child's
performance in the first rhyme detection test given at 314 (at the same time
as the nursery rhyme measure); (6) either the rhyme detection test (417) or
one of the three phoneme detection tasks; (7) the knowledge of nursery
rhyme scores at 314.

Controlling for rhyme. The second part of Table 4 (Controlling for rhyming
skills) gives the three analyses in which the effect of differences in sensitivity
to rhyme at 457 was controlled.
Two points can be made about these analyses. The first is that in all three
the rhyme detection scores in their own right were strong predictors of
reading and spelling one year four months and one year eight months later
even after stringent controls for extraneous factors.
420
TABLE 5. The relation between nursery rhyme knowledge and phoneme detection (six-step fixed-order multiple regressions
and logistic analyses)

Last phoneme-5; 7 First phoneme-5; 7 First phoneme-6;3

Dependent Change in scaled
variable Change in R2 deviation (x2) Change in R2 F Z
a
Step 1 0-03 167 0-03 1 99
RSER

Age at read/spell
test
Step 2 032 2O-39* 041 4401«*
X
I.Q. (W.P.P.S.I.)
Step 3 006 027 0007 076
rMES

Vocab( 3 ;4)
(B.P.V.S.)
Step 4 003 084 002
z
449 059 0
Mothers' educational
level n
Step 5 001 123 048 0002 021 D
Rhyme detection z
(3; 4) 0
Step 6 009 -59 435* 004 388
Nursery rhyme test
(3; 4)

*, p < 005; **, p < 001; *•*, p < 0001.

a
d.f. = 1 for Steps 1—3 and 5-6.
b
d.f. = 4 for Step 4.
 CHILD LANGUAGE

The second point is that when rhyme detection is entered as the sixth step
the relationship between nursery rhyme knowledge and reading and spelling
disappears in all three cases. Thus the children's proficiency in rhyme
detection more than a year after the nursery rhyme test provides a sufficient
explanation for the relationship between nursery rhyme knowledge and
reading and spelling. These results are strong support for the second
prediction and suggest that knowledge of nursery rhymes affects children's
sensitivity to rhyme which in turn has an effect on their success in reading
and spelling.

Controlling for phoneme detection. Table 6 gives the results of seven further
analyses in all of which we controlled for phoneme detection. Here the picture
is not so uniform. The addition of the phoneme detection scores to the
analyses removes the relationship between nursery rhyme knowledge and
reading but it does not have the same effect with spelling. Thus the pathway
between nursery rhymes and reading could be said to be through phono-
logical sensitivity in general since both the rhyme and the phoneme
detection measures account for the relationship between nursery rhymes and
reading. However, the phoneme detection measures on their own do not
account for the relationship between nursery rhymes and spelling since this
still stands when the phoneme detection measures are added. We must
conclude that rhyme plays a special part in the connection between nursery
rhymes and spelling.

Controlling for a non-phonological skill — object naming. It is just possible that

the latter rhyme detection test suppresses the relation between nursery
rhymes and all the reading and spelling tests simply because the rhyme
detection measure was closer in time to the reading and spelling tests than the
nursery rhyme task was. The way to control for this is to see if another, non-
phonological, measure given at the same time as the rhyme detection measure
or later, also removes the connection. Our prediction was that the non-
phonological connection would not have this effect because the pathway
between nursery rhymes and reading is a phonological one. So we ran
multiple regressions and logistic analyses with the same three reading and
spelling dependent variables. The first five steps were the same as those in the
previous analyses. The sixth step was now the object naming test scores and
the seventh the nursery rhyme scores. In all three cases the nursery rhyme
scores were still significant predictors of reading and spelling (Reading 5; 11
f = 385, / X C 0 5 : Spelling 5; 11 £= 13-06, p < o - o o i : Reading 6;3 R2
change = 0-027, (F = 4'57. P < °'°5))- Thus the suppression of the con-
nection between nursery rhyme scores and reading and spelling by the rhyme
detection scores is probably nothing to do with the relative recency of the
rhyme detection test.
422
 NURSERY RHYMES AND READING

TABLE 6. Nursery rhyme knowledge and reading and spelling - phoneme

detection controlled (six- and seven-step fixed-order multiple regressions and
logistic analyses)

Reading-5;11 Spelling-5; 11
Dependent Change in scaled Change in scaled Reading-6; 3
variable deviation (x2) deviation (/ 2 ) Change in R2 F

Opening phoneme detection task (5 ;7) scores controlled

Steps 1-5 As in Table 4
Step 6 2-83 7-14" 006 1060**
Opening phoneme
detection scores
(5; 7)
Step 7 294 603* 0006 113
Nursery rhyme
test (3:4)
End phoneme detection task (5; 7) scores controlled
Steps 1-5 As in Table 4
Step 6 2-73 5-62** 002 236
End phoneme
detection scores
Step 7 206 5-36* 001 225
Nursery rhyme
test (3:4)
Opening phoneme detection task (6; 3) scores controlled
Steps 1-5 — — As in Table 4
Step 6 — — 005 8o6»*
Opening phoneme
detection scores
Step 7 — — 001 209
Nursery rhyme
test (3; 4)

*, p < 0 0 5 ; **, p < 0 0 1 ; **, p < 0001.

d.f. = 1 for Steps 6 & 7 in every case.

Path analysis
The logic of our attempts to establish the pathway between nursery rhyme
knowledge and success in reading and spelling is very similar to the reasoning
used in path analysis. That form of analysis, however, demands continuous
and normally distributed variables, and the fact that several of our in-
dependent and some of our dependent variables were not normally dis-
tributed made it difficult and in most cases impossible for us to use the
technique. However, we did carry out one path analysis with normally
distributed variables.
423
 CHILD LANGUAGE
R2 = 0-595

0158 (n.s.)

Nursery rhyme 0507 Rhyme

knowledge (3 ; 4) / detection
_ (4; 7) Reading (6; 3)
0-388

Fig. i. Path analysis - the indirect route from nursery rhymes to reading. (The numbers in the
figure represent the beta figures in the appropriate multiple regressions.)

The five variables in this analysis were (i) the children's knowledge of
nursery rhymes (3:4); (2) I.Q.; (3) rhyme detection (417); (4) phoneme
detection (6; 3); (5) reading (6; 3). The model that we tested was that the path
from nursery rhyme knowledge to reading would be an indirect one through
rhyme and phoneme detection, and that there would not be a strong direct
path from nursery rhymes to reading.
Figure 1 gives the results of this analysis. The paths from nursery rhyme
knowledge to rhyme and phoneme detection and from those variables
through to reading are very strong. In contrast the direct path from nursery
rhyme knowledge through to reading is the weakest of all and it is not
significant (^(1,59) = 2-005, p = 0-162). Thus the analysis provides striking
support for our idea that nursery rhymes play a role in children's sensitivity
to rhyme and to phonemes and in this way affect children's success in
learning to read.
The analysis supports the idea of a NURSERY RHYME - RHYME DETECTION -
READING route, and a NURSERY RHYME - PHONEME DETECTION - READING route.
It is worth adding that there is a reasonably strong path from nursery rhymes
to phoneme detection here, though this link fell just short of significance in
the earlier multiple regression. This is probably due to the fact that we were
forced to omit some variables from the path analysis (most notably mothers'
educational level) because they were not continuous.
The path analysis also throws up the intriguing possibility of another
route. The pathway between rhyme detection at 457 and phoneme detection
at 6;3 is also strong, and this suggests that rhyme detection itself may lead
to phoneme detection. In other words there may be a continuous de-
velopment of phonological skills starting with sensitivity to rhyme and
developing into sensitivity to phonemes. The path analysis supports the idea
424
 NURSERY RHYMES AND READING

of a NURSERY RHYME - RHYME DETECTION - PHONEME DETECTION - READING

route, in addition to the other two routes that we have already mentioned.

DISCUSSION
The results of this study are particularly exciting because they show a strong
link between an entirely informal experience early on in the child's life and
a formal educational skill which the child must acquire some years later.
When parents introduce their child to nursery rhymes it is most unlikely that
eventual success in reading is in their minds. Yet our study makes it clear that
this early knowledge of nursery rhymes may play a considerable role in
preparing the child for reading and spelling. The fact that there is such a
strong connection even after the most demanding controls have been taken
is obviously important, and our study goes some way towards explaining the
reason for its existence.
Our evidence suggests that the pathway is through the child's growing
sensitivity to the component sounds in words. There are two steps to our
argument. First, the nursery rhyme scores (with one exception) also predict
success in phonological tasks over two years or more and do so even after
their initial levels of phonological sensitivity have been controlled. So
knowledge of nursery rhymes may affect the development of phonological
sensitivity. Of course we need to know a great deal about the way in which
children acquire this knowledge. Trevarthen's work suggests that it may start
early, and that prosodic cues and rhythm in mothers' speech to their babies
may play an important part (Trevarthen & Marwick 1986). This work also
suggests that nursery rhymes are an ingredient of mother-infant ' dialogues'
and thus are a part of the intersubjective routines which may play a
significant role in language acquisition.
Secondly, the relation between nursery rhyme knowledge and reading and
spelling disappears when subsequent rhyme scores, which predict reading
and spelling very well, are entered in the equation. So it is possible that
nursery rhymes enhance children's sensitivity to rhyme and this in turn helps
them to learn to read.
Much the same argument can be made about the phoneme detection tasks
and reading. The relation between nursery rhymes and reading disappears
when phoneme detection scores are entered. We conclude that we have
evidence for a model which takes the following form: nursery rhymes
enhance phonological sensitivity (rhyme and phoneme detection) in general,
which in turn enhance reading.
Spelling is a different matter. The rhyme detection measures account for
the connection between nursery rhymes and spelling, but the phoneme
detection measures do not. Thus rhymes are an essential ingredient of the
connection between nursery rhymes and spelling. At first sight it seems quite
425
 CHILD LANGUAGE

difficult to explain why this should be so. In rhyme the phonological unit is
grosser and less specific than it is in phoneme detection tasks. But we also
have evidence that children with reading problems make better progress in
spelling (though not in reading) if they can rhyme (Bradley & Bryant 1978).
Why should rhyme play such a specific part in the connection with spelling ?
One possibility is that rhyming skills are especially important because they
make it easier for children to learn about sequences of letters and especially
about sequences shared by words which also rhyme. Our idea is that a child
who knows that 'light', 'fight' and 'sight' rhyme is in a better position to
learn about the sound usually associated with '-ight', the spelling pattern
which these words have in common. There is already evidence to support this
suggestion (Bradley 1980, Goswami 1986, Bryant & Goswami 1987, Bradley
1988, Goswami & Bryant 1988). The results that we have just reported
suggest that this sort of learning might be more important for spelling than
for reading.
Whatever the nature of the pathway, it is now quite clear that knowledge
of linguistic routines such as nursery rhymes constitutes an important aspect
of children's linguistic development. We know of no other linguistic variable
that has predicted other aspects of linguistic development over such a long
time and after differences in intelligence, vocabulary, social background and
even the children's initial phonological sensitivity have been controlled. The
strength and specificity of the connection between children's early knowledge
of nursery rhymes and their success in learning about written language some
years later, suggests a causal hypothesis. It is that familiarity with nursery
rhymes enhances children's sensitivity to the component sounds in their
language and that this in turn affects their progress in reading and spelling.
Our longitudinal data are entirely consistent with this causal claim. On their
own, however, they do not completely establish that the claim is correct. We
need evidence as well from intervention studies (Bryant & Bradley 1985). If
our hypothesis is right then extra experience with nursery rhymes should
make children more successful in phonological tasks and it should also help
them to learn to read and spell.
This is the obvious next step in research on this topic. Whatever the
outcome it is already clear from our data that children's early knowledge of
nursery rhymes is strongly connected to other aspects of their linguistic
development.

CONCLUSION
All in all, our results have shown a powerful and lasting connection between
the children's early knowledge of nursery rhymes and aspects of their
linguistic development later on. The nursery rhyme scores are connected to
the development of phonological sensitivity over the next two to three years
426
 NURSERY RHYMES AND READING

and, through that sensitivity, are linked to the children's success in learning
to read and spell as well.

REFERENCES
Bradley, L. (1980). Assessing reading difficulties. Basingstoke and London: Macmillan
Education.
(1988). Making connections in learning to read and to spell. Applied Cognitive Psychology
2. 3-18.
Bradley, L. & Bryant, P. E. (1978). Difficulties in auditory organisation as a possible cause of
reading backwardness. Nature 271. 746-7.
— (1983). Categorising sounds and learning to read - a causal connection. Nature 301.
419-21.
(1985). Rhyme and reason in reading and spelling. I.A.R.L.D. Monographs No. 1.
Ann Arbor: University of Michigan Press.
Bryant, P. E. & Bradley, L. (1985). Children's reading problems. Oxford: Blackwell.
Bryant, P. E. & Goswami, U. (1987). Phonological awareness and learning to read. In J. Beech
and A. Colley (eds), Cognitive approaches to reading. Chichester: John Wiley.
Bruce, D. J. (1964). The analysis of word sounds. British Journal of Educational Psychology
34. 158-70.
Chukovsky, K. (1963). From two to five. Berkeley: University of California Press.
Denckla, M. & Rudel, R. (1976). 'Rapid automatised naming': dyslexia differentiated from
other learning disabilities. Neuropsychologia 14. 471-9.
Glenn, S. M., Cunningham, C. C. & Joyce, P. F. (1981). A study of auditory preferences in
non-handicapped infants and infants with Down's syndrome. Child Development 52.
>303-7-
Goswami, U. (1986). Children's use of analogy in learning to read: a developmental study.
Journal of Experimental Child Psychology 42. 73-83.
Goswami, U. & Bryant, P. E. (1988). Rhyming, analogy and children's reading. In P. B.
Gough (ed.), Reading Acquisition. Hillsdale NJ: Erlbaum.
Knafie, J. D. (1973). Auditory perception of rhyming in kindergarten children. Journal of
Speech and Hearing Research 16. 482-7.
(1974). Children's discrimination of rhyme. Journal of Speech and Hearing Research 17.
367-72.
Lenel, J. C. & Cantor, J. H. (1981). Rhyme recognition and phonemic perception in young
children. Journal of Psycholinguistic Research 10. 57-68.
Liberman, I. Y., Shankweiler, D., Fischer, F. W. & Carter, B. (1974). Explicit syllable and
phoneme segmentation in the young child. Journal of Experimental Child Psychology 18.
201-12.
Liberman, I. Y., Shankweiler, D., Liberman, A. M., Fowler, C. & Fischer, F. W. (1978).
Phonetic segmentation and recoding in the beginning reader. In A. S. Reber and D. L.
Scarborough (eds), Toward a psychology of reading. New York: Erlbaum.
Maclean, M., Bryant, P. E. & Bradley, L. (1987). Rhymes, nursery rhymes and reading in
early childhood. Merrill-Palmer Quarterly 33. 255-82.
Mann, V. (1986). Phonological awareness: the role of reading experience. Cognition 24. 65-92.
Morais, J., Cary, L., Alegria, J. & Bertelson, P. (1979). Does awareness of speech as a
sequence of phones arise spontaneously ? Cognition 7. 323-64.
Morais, J., Bertelson, P., Cary, L. & Alegria, J. (1986). Literacy training and speech
segmentation. Cognition 24. 45-64.
Read, C , Zhang, Y., Nie, H. & Ding, B. (1986). The ability to manipulate speech sounds
depends on knowing alphabetic spelling. Cognition 24. 31-4.
Snow, C , De Blauw, A. & Van Roosmalen, G. (1979). Talking and playing with babies: the
role of ideologies of child-rearing. In M. Bullowa (ed.), Before speech. Cambridge: C.U.P.
427
 CHILD LANGUAGE
Trevarthen, C. (1986). Development of intersubjective motor control in infants. In M. G.
Wade & H. T. A. Whiting (eds), Motor development in children : aspects of coordination and
control. Dordrecht: Martinus Nijhoff.
—— (1987). Sharing makes sense. In R. Steele & T. Treadgold (eds), Language topics - essays
in honour of Michael Halliday. Amsterdam: John Benjamin Publishing Company.
Trevarthen, C. & Marvvick, H. (1986). Signs of motivation for speech in infants, and the
nature of a mother's support for development of language. In B. Lindblom & R.
Zetterstrom (eds), Precursors of early speech. New York: Stockton Press.

428