-
INFORMATION TO USERS
This manuscript has been reproduœd from the microfilm master.
UMI films the
text directiy from the original or copy submitted. Thus, some
thesis and
dissertation copies are in t y w t e r face, while others may be
from any type of
cornputer printer.
The quality of this reproduction is dependent upon the quality
of the copy
submitted. Broken or indistinct pn'nt, colored or poor quality
illustrations and
photographs, ptint bleedthrough, substandard margins, and
improper alignment
can adversely affect reproduction.
In the unlikely event that the author did not send UMI a
camplete manuscript and
there are missing pages, these will be noted. Also, if
unauthorized copyright material had to be removed, a note will
indicate the deletion.
Oversize materials (e.g., maps, drawings, charts) are reproduœd
by sedioning
the original, beginning at the upper lefl-hand corner and
continuing from left to
tight in equal sections with srnall overlaps. Each original is
also photographed in
one exposure and is induded in reduced fom at the back of the
book.
Photographs inciuded in the original manuscript have been
reproduced
xerographically in this copy. Higher quality 6' x 9' black and
white photographie
prints are availaMe for any photographs or illustrations
appearing in this copy for
an additional charge. Contact UMI directiy to order.
Bell 8 Howell Infornation and hamirtg 300 North Zeab Road. Ann
Arbor, MI 48106-1346 USA
600-521-0600
-
PHONOLOGICAL AWARENESS IN CANTONESE ESL STUDENTS
Margaret Yiu-ki Kong Lam
A thesis submitted in confomity with the requirements for the
degree of Master of Arts
Department of Human Development and Applied Psychology Ontario 1
nstitute for Studies in Education
University of Toronto
O Copyright by Margaret Yiu-Ki Kong Lam 1998
-
National Library Bibliothégue nationale du Canada
Acquisitions and Acquisitions et Bibiiographic SeMces seMces
bibliographiques 395 !Uellingtm Street 395, rue Wellingtm OttawaO~
nlAON4 Onnwa ON K1A ON4 canada canada
The author has granted a non- exclusive licence allowing the
National Library of Canada to reproduce, loan, distri'bute or seil
copies of this thesis in miroform, paper or electronic formats.
The author retains ownership of the copyright in this thesis.
Neither the thesis nor substantial extracts fiom it may be printed
or othenivise reproduced without the author's permission.
L'auteur a accordé une licence non exclusive permettant a la
Bibliothèque nationale du Canada de reproduire, prêter, distribuer
ou vendre des copies de cette thèse sous la forme de
microfiche/fih, de reproduction sur papier ou sur format
électronique.
L'auteur conserve la propriété du droit d'auteur qui protège
cette thèse. Ni la thèse ni des extraits substantiels de celle-ci
ne doivent être imprimés ou autrement reproduits sans son
autorisation.
-
PHONOLOGICAL AWARENESS IN CANTONESE ESL STUDENTS
by Margaret Yiu-ki Kong Lam
Master of Arts, 1998 Department of Human Development and Applied
Psychology
Ontario lnstitute for Studies in Education University of
Toronto
ABSTRACT
According to the Contrastive Analysis, Difierential Markedness
and the
Feature Compeütion hypoüieses, specific features disparate
between L1 and L2
may engender negative transfer in L2 leamers, often instantiated
as emrs in their
perception and produdion. In ESL leamers, L I interference would
occur specifically
in acquiring disparate features in L2, or positive transfer
would ocarr for equivalent
and similar features. The resulting developrnental process in L2
phonology in ESL
leamers will thus be disparate from that of L I English
speakers. Diierent
performances on phonological tasks by L I and L2 children at
difierent times in the
prasent study provide support for different developmental
patterns of awareness of
English phonology. Evidence of an L I effect was found in the
positive transfer on
rhyrne discrimination and negative transfer on phoneme
alliteration and blending in
the L2 group. Theoretical implications as well as implications
for programming,
teaching and assessrnent for ESL students are diswssed.
-
ACKNOWLEDGEMENTS
I wish to thank the followi'ng people to whom I am indeMad in
the completion of this
riesis. First of alIl I wish to aiank Dr. Esther Geva, my Thesis
Supervisor and
principle researcher of the ESL At-Risk Project Wmout her
insight and guidance,
and her unyielding support and patience, mis Thesis can never be
cornpleted. I
would also like to thank other members of the research team, in
partiarlar, Robindra
Sidhu and Barbara Schuster for their adviœ, encouragement and
suppoct; and
Luya SegalSeiden and Allison Ma& for advising me on
phonetics and phonology. I
would like also to thank Dr. Dale Willows, member of my Thesis
Supewisory
Cornmittee, for her advice on the content and presentstion
format on the Thesis.
Last but not least, I would like to thank the principals and
staff of al1 the parücipating
schools in York Catholic District School Board for their
cooperation and
encouragement throughout the testing sessions, without their
support, even at the
most taxing moments, this project could never be successful.
iii
-
TABLE OF CONTENTS
.........................................................................................................
ABSTWCT ii ...
................................................................................
ACKNOWLEDGEMENTS I I I
....................................................................................
TABLE OF CONTENTS iv
...............................................................................................
LIST OF TABLES vi
...........................................................................................
LIST OF FIGURES vii
LIST OF APPENDICES
...................................................................................
viii
........................................................................................................
Introduction 1
Theoretical Framework
......................................................................................
2
The Contrastive Analysis Hypothesis
.............................................. 3
The Marked Differential Hypothesis of CAH
..................................... 4
The Feature Cornpetition Model of Segment Transfer
..................... 5
Cross-language Studies of L I -L2 relationship
................................................. 6
Impact of LI on Phonological Awareness of L2
............................... 6
Cross-language Transfer of Phonologid Awareness
..................... 9
Phonological Propetties of Cantonese
...................................................... 10
Developrnent of Phonological Awareness in Cantonese
............... 11
...........................................................................................
The Present Study 13
Research Method
............................................................................................
17
Procedures
......................................................................................
48
Measures
.........................................................................................
18
Results and Discussion
.......................................,..... 23
General Discussion
......................................................................................
35
-
Limitations
.......................................................................................................
43
Relations of Phonological Awareness and Reading and Its
implications ...... 44
References ..................... ..,
....................................................................
60
Appendix A ................................ ..
................................................................
63
Appendix B
......................................................................................................
66
Appendix C
.....................................................................................................
67
Appendix D
.................................................................................................
68
.......................................................................................................
Appendix E 69
-
UST OF TABLES
Table 1.
Table 2.
Table 3.
Table 4.
Table 5.
Table 6.
Table 7.
Table 8.
Table 9.
Table 1 O
Table 12
Mean Scores of Phonoiogics! Tasks by Time
and Language Groups
............................................................... 47
Correlations smong Phonological Tasks-Combined
Groups (Time 1) .............................................
....................... ....... 48 The Effects of Time and Group on
Phonological Tasks:
MANOVA Summary Table
............................................................ 49
Mean Percentage Scores of Auditory Analyk Skills (AAS)
Tasks by Sets ......
......................................................... .........
....... 50 The Effea of Time and Group on Sets of Auditory
Analyüc
Skills Task: MANOVA Summary Table ........ .. ..... ... .. ... .
.. ... . ... ..... .. 51 Mean Scores of Rhyme Discrimination by
Sets, L i and L2 ........ 52 The Effect of Time and Group on sets
of Rhyme
Discrimination: MANOVA Summary Table ....... ...... .... . .....
. ... ..... .. 53 Mean Scores of Word Reading Tasks by Time
and Language Groups
................................................................ 54
Correlation of Reading and Phonological Tasks (lime 3). . . . . . .
. . . .55 Partial Correlation of Combined Mean z-scores of
Phonological and WA and WRAT Tasks, LI and L2,
Controlling for Mean MAT Scores
............................................. 56 Correlation of
Combined Mean z-scores of Phonological and WA and WRAT Tasks, L I
and L2 .. ..... . .. . . .. . . .. 56
-
LIST OF FIGURES
Figure 1 . Comparison of mean percentage score3 of
........................................................................
phonological tasks 57
Figure 2 . Comparison of Rhyme Discrimination. Auditory
Discrimination and Auditory Analytical Skills over time
............... 58
Figure 3 . Comparison of U with L2 Groups on Auditory Analyüc
Skills by Sets
.................................................................................
59
vii
-
LIST OF APPENDICES
Appendix A . Oddity Task Items
.......................................................... 65 .
Appendk B Rhyme Discrimination Task Items
....................................... 66
Appendk C . Auditory Discrimination Task Items
.................................... 67 Appendk 0 . Auditory
Analytic Skills Task Items ......................................
68 Appendix E . Sample of Consent Latter
.................................................... 69
viii
-
INTRODUCTION
"Phonological awareness" refers to the ability to discriminch
and
manipulate sounds in words. There is a convergence of evidence
from
research on reading that phonological awareness is ordinarily
a
prerequisite for reading print (Adams, 1 990; Ehri, 1 995;
Goswami &
Bryant, 1990; Stanovich, 1 992; Swirling & Sternberg, 1 994;
Wagner,
1986). The developmental trajectory of phonological skills in
native
English-speaking children is becoming well establ ished (Adams,
1 990;
Cisero 8 Royer, 1995; Goswami 8 Bryant, 1990). From the data of
a
relatively smaller body of research in second language (L2 )
acquisition,
there is support for a similar pattern in L2. Most
cross-language studies
have compared two alphabetic languages (e.g., English, French,
Czech,
Spanish), whose phonological frameworks are more similar than in
the
case of an alphabetic language and a monosyllabic, tonal
language, such
as Cantonese. Greater disparity in phonology may result in
greater
difficulty for the L2 leamer. Hypotheses such as Contrastive
Analysis,
Markedness and the Feature Cornpetition model are intended to
explain
the roles specific phonological properties of Li may play in the
acquisition
of L2 phonology. Frorn these hypotheses, one cm predict when and
how
-
Cantonese children will experience difficulty in aquiring
English
phonology.
This paper is organized as follows: I will first discuss the
theoretical
framewrk. Second, the findings from cross-language studies
on
phonological awareness are examined to show to what extent they
may
support the hypotheses. A wmparison of the phonological aspects
of
Cantonese and English will be followed by a discussion of
implications of
the empirical data and specific predictions of the difficulties
Cantonese
children may experience in acquiring English. In the following
section, I
will describe the main study and its results as well as the
relevance of the
data to the theoretical framework. Implications for assessrnent
and
educational programrning for ESL students will then be
outlined.
THEORETICAL FRAMEWORK
The hypotheses of Contrastive Anal ysis, Differential
Markedness
and the Feature Competition Model of Segment Transfer
generally
assume an underlying univenal grammar that applies to al1
ianguages.
The "grammat' refers not only to syntactic grammar but
encompasses al1
-
niles goveming a language. Hence, the terni ugrâmmar' is used
here to
refer to phonological niles in a language.
The Contrastive Analysis Hypothesis (CAH)
The Contrastive Analysis Hypothesis (CAH) wes first developed as
a
basis for the preparation of instructiorcal material for
teadiing a second
language (Whitman, 1970). CAH states that one's underlying
knowledge
of LI grammar affects how an individual leams a second language;
the
greater the difference in a specific feature of grammar between
two
languages, the more difficult it is for the leamer to acquire it
in L2; in the
case of disparate features between languages. L 1 interference
wi Il occur.
This version of CAH is sometimes referred to as the 'strong"
version
(Eckman, 1977; Oller & Ziahosseiny, 7970). Howver, as some
CAH
predictions were not corroborated, a "weak" version of CAH
emerged
which attempts to account for rather than specifically predict
the areas of
difficulties: Oller and Ziahosseiny (1 970) proposeci a
'moderate" version
of CAH. This is a modifieci CAH that supports the notion of L
i
interference with an additional corollary to explain errors
derived from
faulty generalization. They suggested that when hm L I -L2
features are
very similar, the leamer wu ld confuse the L2 input because of
the
-
underlying familiarity with L I , resulting in errors of
production. They
found support for their hypothesis in that Spanish students made
more
spelling mors in English than Korean students because of L I
interference.
The Marked Differential Hypothesis of CAH
Eckman (1 977) proposed another modification to CAH by
introducing the hypothesis of Differential Markedness.
Markedness
correlates with 'degree of difficulty" in the grarnmar of a
language.
Degree of diffculty refers to the degree of typological
difference in
features between hrvo languages. New contrasts beniueen two
languages,
for instance, will be of a high degree of difficulty and more
difFicult to leam
than features that are similar. Degrees of difiiculty c m also
be applied
within a language. For instance, voiced stop-consonants (e-g.,
IW, Idl
and I g l ) are more dificult than voiceless ones (e.g., /pl, A/
and lkl).
Markedness refers to the more prominent features and thereby
provides
further explanation for difficulties or m o n that degree of
difficulty alone
cannot explain. Features that are more diffiwlt and more marked
will be
more diffiwlt to aquire.
-
The concept of Markedness benNeen two languages implies that
a
certain feature may be more prominent in one language than in
another
language. If a feature is more marked in L2 than in il, it \MI1
be more
difficult to acquire. However, if a feature is a more marked in
L I than in
L2, it will be conducive for the perception of similar features
in L2. This
argument underlies the hypothesis of feature prominence proposed
by
Hancin-Bhatt (1 994) in her Feature Competition Model.
The Feature Competition Model of Segment Transfer (FCM)
The hypothesis of Feature Competition of Segment Transfer
(FCM)
specifies the condition for interlanguage transfer
(Hencin-Bhatt, 1 994).
The degree of difficulty of acquisition when a feature is to be
transferted
from L I to L2 depends on the disparity of L2 in relation to LI
. Vastly
disparate between-language features will be very diffîw lt to
transfer from
one language to another. Identical between-language features
will
facilitate positive transfer. Similar features will evoke
representations by
mapping on L I grammar, resulting in LI-like sounds in
production.
According to Hancin-Bhatt (1994), these errors wu ld occur until
the
leamer advances to a new stage in which an appropriate constnict
of L2
grammar is built The phenomenon of substitution, in fad, can
be
-
construed according to Hancin-Bhatt, as generalization errors
psited by
the moderate version of CAH.
Apart from beheen-language differences, the influence of
"prorninent' features in a language also accounts for perception
cif L2
input. Prominence refers to the weight a feature demands in
the
individual's perception. While there may be a universal rule of
feature
prominence across different languages (e.g., nasality is a more
prominent
feature across languages), ( Hancin-B hatt, 1 994) t here are
also language-
specific prominent features. For each language, therefore, a
hierarchy of
feature prominence can be constnicted for native speakers.
More
prominent features in a L1 language system will be more salient
in the
leamers' perception, and therefore, conducive to the perception
of similar
features in 12. On the other hand, less prominent L I features
will escape
the leamer's attention, and therefore, not easily acquired in
the L2.
CROSS-LANGUAGE STUDIES OF L I 4 2 RELATIONSHI?
The Impact of L I on Phonological Awareness of L2
Evidence of the influence of L I on performance in
phonological
tasks in L2 was found in a number of cross-language studies.
Cavavolas
-
and Bruck (1993) compared CZ8ch and English children and foücd
that
Czech cnildren wera better in discriminating consonant clusten
in the
onset position than the English speaking children. They argued
that
consonant clusters are more prominent in the Czech language. On
the
other hand, English children did better in discriminating single
onset
consonants, arguably because it is a feature more salient in
English. The
samples cited by Watson (1991 ) clearly indicate the effect of L
I in young
children's utterance in L2. Cisero and Royer (1 995) found, in
their study
of English- and Spanish- speaking diildren, that Spanish
children made
more errors on English tasks involving phonemes than English as
LI
children. Although the authors concluded that cross-language
transfer of
phonological perception is present, the effect of L i on the
perception of
L2 phonology was evident.
Huang and Henley (1 995) attributed the difference in
performance in
phonemic deletion tasks for Chinese and English children to the
different
phonemic structures of the tw languages. Studies of
French-speaking
students found that they do better in phonological tasks
involving
syllables than their English-speaking munterparts (Bnick &
Genesee,
1995; Cutler, Mehler, Norris & Segui, 1992). Bru& and
Genesee (1 995)
reported that kindergarten English-French bilingual children
excel in
-
segmentation of syllables as compared to English monolingual
children,
and Cutiet et ai. (1992) found similar tendencies in French
dominant
adults. The differences in performance in both cases w r e
attributed to
the more prominent feature of syllables in the French language.
The
language-specific features showed significant effects in the
phonological
tasks and thereby provide support to the FCM hypothesis.
However, the
difference in performance in Bruck and Genesee's sample
disappeared
one year later when the children were exposed to reading
instruction in
Grade One. Similarly, Fashola (1 997) found that Spanish
speaking
children made more spelling errors in English words than
English-
speaking children in Grades Two and Three, indicating that
children
employ L I phonological knowledge in the early stages of leaming
to read
in L2. She also found that error rates that can be explained by
an L I
effect dissipated by Grade Four. Again the evidence clearly
supports the
Feature Cornpetition Model as well as the Marked Differential
version of
the Contrastive Analysis Hypothesis in dernonstrating a negative
transfer
of the knowledge of L1 phonology in the acquisition of L2.
Further
support to FCM is shown in the amelioration of the L i effect
over time.
The phenornenon may testfy to the fact that with instruction,
the L2
-
phonological construct is developed gradually and errors becorne
few and
far between.
Cross Language Transfer of Phonological Awareness
A number of cross-language investigations involving L I -L2
have
reported clear evidence of interlanguage transfer in children
who are
aquiring a second language (Cisero 8 Royer, 1995; Durgunoglu,
Nagy 8
Hancin, 1993; Snow, 1992; Verheoven, 1994). In these
studies,
difFerences in performance in phonological tasks have often been
found.
In some cases, transfer is evident for some tasks but not others
(e.g.,
Verheoven, 1994). ln these cases, the influence of L I grammar
cannot
be ruled out. These studies involved English and alphabetical
languages
such as Spanish, French, and Turkish; there are few studies
(e.g., Huang
& Henley, 1995) involving a non-alphabetic language. We
shall assume
similar pattern of transference is tnie between Cantonese (LI)
and
English (L2) for purposes of analysis in the present study.
-
PHONOLOGCAL PROPERTES OF CANTONESE
Since the present investigation deals with Cantonese
children's
acquisition of phonological awareness in English, it is
necessary, at this
point, to compare briefly the phonological properties of
Cantonese and
English. Cantonese is a dialect dominant in over thirty million
people in
South China, in the Quangtung Province, and in Hong Kong.
Immigrants
from these areas have settled in North America and elsewhere in
the
world (So & Dodd, 1995). Cantonese is a monosyllabic, tonal
and
morphernic language. Each syllable represents a morpheme which
can
be produced in written fom as a character in Chinese. A
correspondence
to an English word rnay consist of one, two or several
mono-syllabic
characters in Cantonese. The written form of Cantonese employs
the
same charaden as Mandarin. For the purpose of this study, I will
refer to
Cantonese as the spoken language. Each Cantonese syllable c m
be
pronounced in nine tcnes; each tom represents a morpheme. The
tones
can be described as rising, lave1 and falling tones of high,
middle and low
pitcb (So, 1989; So 8 Dodd, 1995). Morphemes derived from
different
tones of the same syllable do not necessary relate to each other
in
meaning. Tones in Cantonese are less marked and carry a
heavier
-
functional load than articulation of consonants (So & Dodd,
1995). It
should be noted that stress is a more prominent feature in
English as
compared to French or Spanish in which syllable are more
prominent
(Bruck & Genesee, 1995; Cutler et al., 1992; Durgunoglu et
al., 1991).
In Cantonese, there are seventeen initial consonants and two
initial
consonant clusters, whereas in Eng lish there are twenty-three
and twenty-
one, respectively. In the final position, Cantonese has eight
consonants
Mi le English has twenty-one. Cantonese has eight vowel sounds,
but
English has twenty-one. The hm consonant clusters in Cantonese
(e.g.,
lkwl and lkhwl) are often spoken in reduced fom by young
children,
indicating the comparative markedness of clusters in the
language (So 8
Dodd, 1995).
Development of Phonological Awareness in Cantonese
Studies on the developmental trajectory of phonological
awareness
in English as L1 reading suggest that children progress from
the
awareness of syllables, to awareness of onsets and rimes, and
then
individual phonemes in initial, middle and final positions in
words ( Cisero
& Royer, 1995; Goswami, 1 990; Goswami & Bryant, 1990;
Stanovich,
1992; Wagner et. al.; 1993). There is evidence that Cantonese
children's
-
development of phonological skills also follows a similar
pattern (So à
Dodd, 1995). By age five, children already have a
weiideveloped
phonological system in their native language. When children
begin to
leam a second language at this age, they need to develop a
new
phonological systern. As is evident from the discussion in
previous
sections, both the CAH and FCM hypotheses predict that the
phonological nature of the L I would influence the acquisition
of the L2
phonological system. If these hypotheses are valid, then it may
be
hypothesized that the more phonologically different hvo
languages are,
the more difficult wuld the acquisition of these features in L2
be.
In their study of development of phonological skills of
Cantonese
children, So and Dodd (1 995) found the majority of their
subjects had
complete mastery of al1 nine tones at age two. Mastery of
final
consonants and vowel sounds was also cornmon in most of their
sample
by that age. Far more children made phonological erron than
tonal
errors, which were close to negligible (.7%). On the consonant
sounds,
Cantonese children could master voiceless stops (e.g., /pl, lu),
and nasals
( e.g., /mJ, in/) before a glide (e.g., Ili). There are no
voiced stopsin
Cantonese. In general, the development of consonant sounds
in
Cantonese children was similar to that of their English
wunterparts. The
-
development of awareness of initial consonants in Cantonese
chiidren
can be demonstrated in children at ages hm, three ard four among
their
participants. At age Mm, a child had a repertoire of ten
initial
consonants, whereas the child at age three had fifteen; the one
at age
four reached wmplete mastery of al1 consonant sounds (i.e., 17).
English
children would nomally attain wmplete rnastery of al1 consonant
sounds
by age four and a half.
Given the phonological disparity betvireen Cantonese and
English,
the CAH and FCM hypotheses may be applied to predict that
specific
errors will occur in the perception and production of English by
Cantonese
leamers. These errors would presumably refiect the specific
properties in
Cantonese phonological grammar and would occur more frequently
in the
early stages of children's leaming to speak and read in that
language
(e.g., in Grade One).
THE PRESENT STUDY
The present study is part of a larger project Mich
investigates
normal and at-risk oral and literacy development in ESL
children. The
present study investigates the developrnent of phonological
awareness of
-
Engiish as L2 in Cantonese immigrant children as they began
English as
a second language (ESL) instruction at school. My objectives
were
a) to compare the deveiopmental trajectories of English
phanological
awareness of Cantonese children with the trajectories of
English-
speaking children for common and disparate features; and
b) to investigate the extent to which different trends in
development may
be associated with the disparate features in phonology between
the
hrvo languages.
The following hypotheses were proposed:
1. Pefitmance of Cantonese children in phonological fasks, like
that of
their English-speaking counterpatts, wouM impmve over time.
As
show in some studies, an LI effed is more prominent in the
early
stages of leaming English (Bnrck & Genesee, 1995, Fashola,
1997),
thus, it was expected that enors made by Cantonese children
wuld
decrease with time. If the same phonological tasks are
administered
across a span of time, exposure and instruction in English
reading will
improve Cantonese children's phonological awareness.
Convergent
ta experimental findings, the performance of Cantonese children
on al!
Eng lish-related phonological tasks will likewise improve over
time.
-
2. Developmental trajixtofies of Englsh phonobgical awareness
of
Cantonese chiJdren will be disparate fmm t h e of their L I pars
.
Congruent with the previous hypothesis, Cantonese children will
not
only likely start at a lower point in knowledge of the English
phonolqy
because of a la& of exposure to the language but also corne
with a
knowledge of different phonological concepts in their L I which
will
likely affect their perception of the phonological features in
the L2
(Geva, 1997). With instruction and increased exposure to the
language, they will make sufficient gains and attain mastery
which
may eventually match thet of their L I peers, despite the
dissimilarity of
the respective developmental trajectories.
3. Cantonese children wi/l show negative transfer when leaming
the
phonologica/î'y more disparate features in L2 . Pursuant to
Hypothesis
2, although Cantonese children should make progress over time
in
their acquisition of the Engl ish phonolog y, particular
features wil l
remain more difficult to master than others: For example,
features
such as consonant clusters in English will be more marked to
Cantonese children and will carry less weight in their
perception.
According to CAH, L I interference will occur in the acquisition
of these
-
features. Similarly, according to FCM, L I influence will be
show in
inaccurate perception and production of consonant clusters.
4. Cantonese children will show positive transfer in leaming
phonologicaIly similar between-language features. For exam p
le,
voiceless stops (e.g., /pl, IU and Ad) and nasal consonants
(e.g., /ml,
hl and h g / ) are present both in English and in Cantonese.
According
to FCM, equivalent features will be easily transferable.
Therefore,
Cantonese children will perform as wll as English children
because of
positive transfer.
5. Cantonese children will show positive transfer in leaming
features in
English that are more prominent in their LI. As noted above,
one
striking disparity between Cantonese and English is the
prominence of
tones in the former. Tones and syllables are less marked in
Cantonese, as they are mastered at an early age (So, 1989),
whereas
syllable is a more marked feature in English than in French or
Spanish
(Bnick & Genesee, 1995, Cutler et Al., 1993). Cantonese
syllables are
mostly of the C-VC structure (e.g., bah/ meaning "fathet), and
the V-
C structure (e.g., lyil, meaning two). According to the
hypothesis of
Differenüal Markedness, the rime (e.g., the /ah/ part of the
syllable
Ibahl) will be less marked than the onset (e-g., the Ibl- part
of the
-
syllable). As the tom in Cantonese is part and parcei of the
syllable,
awareness of tones wili entail better awareness of the syllable
for the
Cantonese cklldren.
RESEARCH METHOD
The study involved 34 Cantonese-speaking participants (1 6
boys
and 18 girls), in grade 1 from 4 schools in hm suburban areas of
a
metropolitan city in southem Ontario, Canada. There were 36
English L I
children (20 boys and 15 girls) from 4 schools in another
suburban area
of the sarne city participating in the same project as the
cornparison
group. A survey of al1 the Cantonese participants indicated that
al1 except
one attend additional literacy instructions of their native
language on
weekends. Written consent from the parents was obtained from
al1
parents whose children participated in the study. The mean age
of the
Cantonese children at the beginning of the project was 75.2
months, and
the mean age of the cornparison group was 76.7 months.
-
PROCEDURE
The children were tested in the fall (Time 1 ) and spring (Time
2) of
their grade 1 year and again in the fall (Time 3) of their grade
2 year. All
testing sessions were conducted individually in English within
the
children's schools in a quiet room. In the first two sessions,
however,
Cantonese translations of instructions w r e provided by a
bilingual
researcher to the Cantonese participants when necessary.
MEASURES
Cognitive tasks
The cognitive task chosen for the present study was the Digit
Span
subtest of the Weschler Intelligence Scale for Children-Third
Edition
(WISC-III) and the Matrix Analogy Test (MAT). The rationale for
including
a cognitive test is based on evidence gathered ftom previous
research
indicating that cognitive abilities of children significantly
predids their
ability to perfom phonological tasks (Wagner, Torgesen,
Laughon,
Simmons, & Rashotte, 1993).
Language task
The Peaboày Picfure Vocabulary Test-Revised (PPVT-R), a test
of
receptive English vocabulary, was used as a language
proficiency
-
measure. Because of a second-language factor, the PPVT-R is not
a
valid masure of verbal ability for the Cantonese
participants.
Phonological Tasks
The phonological tasks in this study w r e chosen with a fows
on
testing the initial stages in the developmental level of
phonological
awareness expected in the early stage of reading, as established
by
previous research studies (Cisero & Royer, 1995; Goswami
& Bryant,
1990; Wagner & Torgesen, 1987). In some tasks, items were
adapted to
the disparate phonology between Cantonese and English in order
to
compare the two language groups on awareness of specific
speech
sounds. Each of these tasks is described below.
Oddity Task
The recognition of onset consonants has been shown to be the
first
skill aquired among young children (Bryant & Bradley, 1983;
Cisero &
Royer, 1995; Kirtley et al., 1989; Treiman 8 Zukowski, 1996;
Wagner et
al., 1993). Two target sounds ( Le., JsW and ItW) w r e chosen
among the
many disparate consonant sounds between the hrvo languages.
The
phoneme 'lsW was contrasted with '/sr, "W, 'IzP, 'lvP and 'M",
and
"IW was wntrasted with "/v/" and '/A''. The children listened to
3-item
sets of auditory stimuli consisting of three pseudowords (e.g.,
SHAT,
-
SHAM, SAN). At the same time, they were presented with three
wooden
chips representing the three sounds. They were asked to point to
the
chip that had a different beginning sound (e.g., SAN). Children
were
given practice items to ensure comprehension of the task demands
before
test items were administered. The wooden chips were used to
avoid an
unnecessary load on memory and to eliminate production related
errors.
Pseudowords were used instead of real words to avoid a
familiarity effect
which might favour L i children. To ensure al1 participants
received the
same stimuli, al1 items was tape-recorded. Each target
phoneme
appeared six times. The two sets of target sounds were mixed
randomly
with a set of six control sounds common !O both languages and
appearing
in various positions within each set.
Rhyme Discrimination (RD) Task
There is abundant research support indicating that the awareness
of
rhymes is a skill mastered by most beginning readers (Bryant
& Bradley,
1983; Cisero 8 Royer, 1 995; Kirtley et al., 1989; Treiman &
Zukowski,
1996; Wagner et al., 1993). A rhyme discrimination task was
designed for
this study with the same target sounds as in the Oddity task
(i-e., IsN and
Ithl). The sounds were also contrasted in the same way as in
the
previous task. The children are given a target pseudoword (e.g.,
TESH),
-
and then a set of three words (e.g., SESH, MES, NEF). The child
is asked
to point to the a number presented on a template to indicate the
w r a that
rhymes with the target word. Again, each target phoneme appeared
6
times. The phonemes are mked randornly with a set of six control
sounds
that are common to both languages. As is the case with
Oddity,
verbalkation is discouraged. The children are given practice
items
delivered by a puppet in order to ensure that they understand
the task
demands.
Auditory Discrimination (AD) Task
An experimental task similar to the Auditory Discrimination
Test
(ADT) (Wepman, 1973) using 34 pairs of pseudowords was used
(e.g.,
thop). The task consists of English pseudowrds with phonemes
that are
absent from the Cantonese language. Like the ADT, these
pseudoword
pairs target phonemes in different positions, including both the
onset and
final positions of words. Of the 34 pairs of items, 14 are
control pairs
where the answer is "same". On the remaining 20 pairs, the
correct
answer is "different". In order to ensure al1 participants heard
the same
sounds, the Oddity, RD and AD tasks wwe tapeiecorded in a
professional studio, with a female voice.
-
Auditory Analytical Skills (AAS) Task
A task adapted from the Test of Auditory Analytical Skills
(Rosner,
1975) was developed to test children's ability to manipulate
phonemes. In
order to minimize the effect of lexical familiarity which might
advantage L i
children, the target wrds, as well as the product of blending
after
al1 iteration of syllables or phonemes, w r e al l real high
frequency Eng lish
words (e.g., belt, bet). The task consists of six sections of
progressive
difficulty, i.e., stripping syllables from words and stripping
phonemes in
words. Again the positions of phonemes being deleted were
arranged
with progressive difficulty according to established
developmental trends,
Le., from the initial to final consonants, from single consonant
to
consonant clusters. Children were given practice items before
the actual
administration of the test.
Reading Tasks
Three different tasks were included. The first was the Word
Attack
(WA) subtest from the Woodcock Reading Mastery Tests,
Revised
Edition (WRMT-R) (Woodcock, 1972). A second measure for
reading
was the Reading subtest of the Wide Range Reading Test,
Revised
Edition (WRAT-R). The third reading measure was an
experirnental
wrdieading task (DR) which wnsisted of 16 frequently used words
at
-
the grade 1 level (9.g.. dogs, thick). The WA task was selrcted
because
it consists of phonetically regular non-words. Since L2 children
may have
a lower knowledge of English vocabulary, the use of non-wrds may
avoid
the confound of prior knowledge which may affect scores as 11
children
rnight read the visual representation of "sight words* rather
than sounding
out letters. The WRAT-R w r d reading was selected because it
provides
a standardized measure of the reading ability relevant to the
current
levels of reading. The B R supplied another measure of isolated
w r d
reading and which included only high frequency wrds.
RESULTS AND DISCUSSION
Before proceeding to examine the results on phonological
tasks,
mean scores on measures of memory and language were analyzed
by
language groups. Analysis of variance on mean scores on the
Digit Span
did not reveal a significant difference (F (1,67)= 2.524, p>
0. l), indicating
that the hivo groups did not differ in rote and wrking memory
abilities.
However, analysis of variance on the MAT scores demonstrated
a
signifiant difference (F(1,67)= 26.13, pe.001). This result
suggests that
the L2 children had significantly higher visual-spatial
reasoning and non-
-
verbal problem-solving skills than the L1 participarts. On the
other hand,
on the language measure, the PPVT-R, analysis of variance found
a
significant difference (F (1, 67) = 21.05, p
-
percentage correct responses on the four phonological tasks
between
groups and times is depicted graphically in Figure 1. Scores on
the AD
reached close to or above 80% at Times 2 and 3 in boiJi
groups,
indicating that the task was relatively easy for the children.
This result
was perhaps due to the fact that although Oddity and RD w r e
also
similar tasks, the AD task presented tm aural stimuli for
discrimination
whereas the other two tasks had three stimuli presented each
time.
Conversely, for the AAS, a task requiring children to
distinguish and
manipulate individual sounds and phonemes, scores appear to
be
considerably lower than for the auditory discrimination tasks.
While it is
possible that in general, phoneme production is more difficult
than
auditory discrimination for children, the difference in
performance may
also be explained in developmental ternis. Thus, these results
provide
some support for the contention that children develop the
ability to
segment and manipulate phonernes Mer they begin to read,
usually
behrveen grades 1 and 2 (Adam, 1990, Goswami 8 Bryant,
1990).
Correlations at Time 1 among individual phonological tasks for
the
hnio language groups combined are relatively moderate (se8 Table
2).
The correlation behNeen Oddity and AAS for the hm groups
combined is
the highest at r=.44; (p
-
that a considerable portion of the phonological tasks for the
participating
children as a group are unrelated and probably measure
differe~t
abilities. However, a look into the correlations among phonolog
ical tasks
within each language groups (see also Table 2) indicates that
the tasks
are likely less related to one another for the L2 group ( r=
0.21 to 0.38)
than in the L7 (r= 0.24 to 0.64). It is probably that at least
at Time 1, the
L2 children w r e less familiar with the English phonological
features in
these tasks than the L I group.
Cornparison of the mean scores on ail the phonological tasks
over
time was performed by multiple analysis of variance (see Table
3).
Robust time effects are seen on al1 four tasks, supporting the
hypothesis
that the phonological awareness of Cantonese children and their
L I peers
as a group irnproved significantly over time. By tirne 3, the L2
children's
performance on most phonological tasks was close, or even
exceeded the
level of their L i peers. It is noteworthy that the L2 children
appeared to
achieve higher scores across the three times on both Oddity and
Rhyme
Discrimination, indicating that there is Iikely a positive
transfer in these
phonological features from their L1.
A cornparison of two times (i.e., Times 1 and 2; and Times 2 and
3)
across the four phonological tasks indicates that the time
effects w r e
-
mostly significant between Times 1 and 2, but not between Times
2 and 3,
with Oddity being the only exception (see also, Table 3). This
indicates
that as a group, the children in both language groups gained
progress
mostly betwen times 1 and 2, i.e., within their grade 1 school
year.
Turning to the interaction between time and group on the four
tasks
(see also Table 3), more robust support for the hypothesis that
the L2
children aquire phonological awareness at a different rate from
their L I
peers is evident. Significant interactions between language
groups and
time were obtained for Rhyme Discrimination (RD) (F(2, 59)=
3.601, pc
.05), Auditory Discrimination (AD) (F (2, 59)= 12.251, p<
.001) and AAS
(F (2, 59) = 6.460, pc .01). On the other hand, the interaction
was not
significant for the Oddity task (F (2, 59)= .030, p> 0.9).
These data
suggest that the growth in phonological awareness of the
Cantonese
children was indeed on a disparate trajectory from the one for
native
speakers of English.
Different rates of growth w r e even more apparent in comparing
the
time and language group interaction between Times 1 and 2 and
between
Times 2 and 3. Hawever, different growth trajectories are seen
in
different tasks. On Rhyme Discrimination (RD), a significant
interaction
was present between Times 1 and 2 (F (1, 57) = 4.857, pe.05) but
not
-
between Times 2 and 3 (F (1,57)= 2.1 04, p> 0.1 ). On
Auditory
Discrimination (AD), a task consisting of word pairs with onsets
and final
consonants randomly inter-mixed, there was a significant
interaction on
both occasions. However, the interaction between Times 1 and 2
was
more robust (F (1, 59)= 8.62, p< .006) than between Times 2
and 3 (F (1,
59)= 5.72, p< .05). On the other hand, for AAS, a task
requiring
manipulation of phonemes, the time and group interaction between
times
2 and 3 was robust (F ( i , 57 ) = 19.22, p< .O01 ) whereas
there was no
between-groups difference between Times 1 and 2 (F (1,57)= 1.1
3, p>
0.2). These significantly disparate developmental trajectories
for RD, AD
and AAS tasks are depicted visually in Figure 2.
Further evidenœ of disparate growth patterns in phonologid
awareness between the L I and L2 groups is shown when different
items
in the AAS task are analyzed in groups of increasing diffiwlty.
Since the
task required children to strip a phoneme from a given word and
blend the
remaining phonemes to make another word, it is a more demanding
task.
Degree of dificulty on this task is arranged by demanding the
alliteration
of phonemes of different combinations placed in difhrent
positions in
words. By grouping items according to the degree of diffiailty,
children's
performance would indicate the extent of their awareness of
the
-
phonology in the English wrd. Items on the AAS can be
categorized into
six sets. Set 1 involves the alliteration of entire syllables,
and sets 2 to 6
require the alliteration of individual phonemes and blending the
rernainirig
phonemes into words. Sets 2 and 3 involve the alliteration of
single
consonants in words; for Set 2, the initial consonants, and Set
3, the final
consonant. Sets 4 to 6 involve the manipulation of consonant
blends in
either the onset or the final positions of words. Set 4 requires
the
alliteration of the first consonant in an initial blend; Set 5,
the first
consonant in a final blend; and Set 6, the second consonant in
an initial
blend.
Percentage mean scores by group on the AAS by sets are show
in
Table 4. In general, both groups of chiidren did well on set 1,
and they
have reached the ceiling of correct responses across three
times. For
Sets 2, 3 and 4, however, disparity in performance between
groups was
apparent. In these three sets, the L I group performed far
better than L2
children at times 1 and 2. At time 3, however, 12 children's
performance
appeared to at least catch up to the L i group. Multiple
analysis of
variance for al1 six sets over time was perfomed (see Table 5).
For sets
2, 3 and 4 significant interactions were obtained between times
2 and 3
but not for tirne 1 and 2. Depicted visually, the mean scores
for these
-
sets demonstrates that the growth over time for the L'l group
indicated a
steady growth rate across tirne, while for the L2 group, much of
the growth
appeared between Times 2 and 3 (see Figure 3). For set 5, task
requiring
the alliteration of the first wnsonant in a final blend, and for
set 6 in which
the alliteration of a second wnsonant frorn an initial blend was
required,
the L2 children did surptisingly better than the LI group at
Times 1 and 2.
It is probable that generally speaking, both of these tasks are
difficult for
the English as L I children, and hence, they achieved low scores
on both
tasks. On the other hand, scores of the L2 children on Sets 5
and 6 at
Times 1 and 2 were similar to those for set 4, which also
involved
consonant blends. It is likely that task demands on sets 5 and
6, though
appeared more difîlwlt for the English as L I children, wre
similarly novel
features for the L2 children. It is also possible that the L2
participants in
the present study had higher decoding skills as show in their
higher
scores on w r d reading tasks (see Table 8). Congruent with the
pattern
of g r o M in set 4, the L2 group showed spurts of growth
b€3ONe8fl Times 2
and 3 on both sets 5 and 6. Conversely, growth of the L1 group
on these
tasks during the same time period was minimal, suggesting that
even in
grade 2, consonant blends presented a challenge to the English
as Li
children.
-
Items on the Rhyme Discrimination task were also ~ a t e g ~ f i
~ e d into
three sets and analyzed. Set 1 consists of cornparison sounds
(e.g., lapl)
at the final position, Mile Set 2 targets the Ishl and Set 3,
the Ithl
phoneme. Mean scores for the thm8 sets are listed in Table 6.
Set 3,
which targets the /th/ phoneme appeared to be difficult for both
groups.
No significant gain was apparent for both groups at time 3 for
this set of
items. For the other two sets, the performance of the L I and L2
groups
was close although, by Time 3, the L2 group made more gains
and
exceeded the L I group. Multiple analysis of variance of time
and group
effects on these sets (see Table 7) indicated that both groups
made
signifiant gains beONeen Times 1 and 2 on al1 three sets, while
the most
apparent gain was show in set 2 which targeted the phoneme /SN.
The
only significant interaction betmen group and time was found for
Set 2
between Times 2 and 3 (F (1, 57) = 4.08, p< 0.05), indicating
a different
growth rate betvveen the Mm language groups in disairninating
this
phoneme at the final position.
Mean scores of al1 three word reading tasks by time and groups
are
listed in Table 8. As can be seen at Time 1, rnean scores on the
WRAT
and EXR tasks differed significantly between the two groups (F
(1,59) =
5.46 and 4.41, pc 0.05, respectively) in favour of the Cantonese
as LI
-
children. By Time 3, th8 L2 group had made even more gains on
WRAT
word reading task than their L1 counierparts (F (1, 59) = 5.45,
pc 0.05).
On the other hand, for the Word Attack (WA) task, there was
no
significant difference beîween the two groups at al1 three
times.
Correlation of reading tasks with phonological tasks in al1
three testing
sessions for Time 2 for the L I and L2 groups combined as well
as for
individual groups are summarized in Table 9. Correlations of
the
individual phonological tasks with each reading tasks indicated
that they
wre signifimntly correlated with w r d reading tasks for both
language
groups. It is also tnie that for both groups, correlations of
Word Attack
(WA), a task of reading non-words with al1 four phonological
tasks are
similarly as signifiant as those for WRAT and EXR which
involved
reading real wrds. While al1 non-words of the WA task were
phonetically
regular, some of the real wrds on the other two word reading
tasks were
phoneticall y irregular. Arnong the four phonological tasks, the
Auditory
Analytic Skills (AAS) task has the highest correlation with al1
three
reading tasks for two groups combined (r=.61 to .66, pe.001) as
well as
for the L1 gnxip (r= .61 to 63 , p< 0.001). However, for the
L2 group the
correlations b e h e n AAS and WA and WRAT w r e sornewhat
higher ( r=
.70 and -71 p
-
.O01 ) whereas the correlations between reading tasks ana AC
were lower
(r=.39 to .44) than for their L I counterparts (r= .53 to .57).
It is possible
that a task involving auditory discrimination is not as robust a
tesi for the
ability to manipulate phonemes as AAS. At the same time, the AD
task
involved vowels that were generally more difficult for the L2
children.
Compared to their Li peers, the ability to manipulate phonemes
appears
to be more closely related to basic w r d recognition and
decoding skills in
L2 children.
In order to illustrate the variance in reading that can be
accounted
for by phonological awareness, partial correlations were
perfomed on
cornbined mean z scores on al1 four phonological tasks and two
reading
tasks for Times 1, 2 and 3 Mile controlling for the variance due
to MAT, a
measure for non-verbal intelligence (see Table 10). Due to a
technical
error, scores on the EXR w r e not available for al1 children at
Time 3.
The cornbined z-scores were based, therefore, only on the WRAT
and
WA tasks. As cm be seen in Table 10, the partial correlation
among
phonological and reading tasks are significant throughout Times
1, 2 and
3 for both language groups. These results indicate that
phonological
processing skills account for approximately half of the variance
( from
41 % to 67%) in word reading tasks for the participating English
as LI
-
children. On the other hand, phonological awareness accounts
for
approximately 30% of the variance (Le., from 26% to 34%) for the
L2
group. In congruence vuith results from a number of previous
studies on
the relation of phonological awareness and reading (Goswami 8
Bryant,
1990; Wagner et al., 1996), phonological awareness continues to
account
for variance of w r d reading over time for the English L I
children and
Cantonese children. It is noteworthy that for the L2 group,
phonological
awareness accounted for approximately similar portions of
variance of
word reading over time. Comparing the partial correlations show
in
Table 10 with the correlations among phonological and reading
tasks for
which the variance due to MAT was not controlled for (as shown
in Table
1 1 ) demonstrate the extent to which variance of word reading
measures
that can be attributed to intelligence. There are no significant
difference
among the two correlations for the L i group, suggesting that
for these
children, MAT scores probably did not affect performance on
reading. On
the other hand, for the L2 children, there is close to 10% of
the variance
of reading rneasures that can be attributed to intelligence,
indicating that,
in the present case, non-verbal reasoning abilities played a
considerable
role on the leaming of a second language.
-
GENERAL DISCUSSION
Robust time effect on al1 phonological measures for both
language
groups provide significant support for the hypothesis that the
L2 children,
like their L I peers, improve in their performance on various
aspects of
phonological awareness over time. This finding suggests that al1
chi ldren
benefit from instruction and exposure to English reading
material,
irrespective of their native language.
The presence of a second language effect was evident in the
significantly different vocabulary scores for the two language
groups
throughout Times 1, 2 and 3, although, the differences between
groups
gradually diminished over time. Results from the present study
appear to
confimi the presence of an L1 effect. As was hypothesized in
the
Contrastive Analysis Hypothesis (Eckrnan, 1977; Whitman, 1 970)
and the
Feature Cornpetition Model of Segment Transfer (Hancin-Bhatt,
1994).
the phonological characteristics of the native language play a
part in the
leaming of aie L2. Firstly, compared to their LI peers,
performance on
phonological tasks suggest that Cantonese children approach
phonological tasks in a somewhat different manner: the
specific
intercorrelations of phonological tasks at Time 1 for the
Cantonese group
-
w r e not as high as those for the L I group. At least in the
beginning
stages of leaming to read, Cantonese children appeared to
treat
individual phonemes such as initial consonant sounds as
independent
from other phonernes (e.g., rhymes). As well, unlike their L i
peers,
Cantonese children performed equally well on tasks involving
consonant
blends, irrespective of their positions in a word. These
findings indicate
that the L2 children probably had a different underlying
knowîedge of
English phonology from that of the L I group.
Perhaps due to the presence of a disparate knowîedge of the
English phonology at the initial stage, developmental
trajectories for the
two language groups are quite distinct as show by the
significant
interaction of time and language groups on three tasks, Le.,
Rhyme
Discrimination, Auditory Discrimination and Auditory Analytic
Skills tasks.
As discussed earlier, awareness of onset consonants is a skill
acquired at
the early stages of reading (Adams, 1990; Goswami 8 Bryant,
1990); the
task of discrimination of onset phonemes is likely equally easy
for both
groups of children. For both language groups, however,
significantly
dîfFerent rates of growth wre obsenred between Times 1 and 2 on
Rhyme
Discrimination, an apparently easier task, whereas on a more
difficult task
such as phoneme alliteration and blending (i.e., Auditory
Analytic Skills
-
task), different rates of growth occurred between Times 2 and 3.
On the
other hand, probabiy due to a mixed degree of difficulty of the
positions of
the target phonemes on the Auditory Discrimination task,
different rates of
growth occurred throughout Times 1, 2 and 3 for various item
types. As
wll, although the Cantonese children achieved lower scores on
more
demanding phonological tasks such as Auditory Discrimination
(AD) and
Auditory Analytic Skills (AAS) tasks at Times 1 and 2, their
significant
gains were apparent on al1 phonological measures at Time 3. In
other
words, the developmental trajectories of Cantonese children on
these
tasks reflected quadratic functions. On the other hand, the L1
group
showed relatively steady lower growth in skills as evidenced in
their
developmental trajectories in phonological awareness on these
huo tasks.
Even though both groups made significant gains by Time 3, it is
worth
noting that by Time 3, the L2 children not only acquired skills
to match
their L I pers but also exceeded the Li children on most
phonological
measures.
Evidence for the presence of an LI interference effect
causing
negative transfer as hypothesized by the CAH (Whitman, 1970) and
the
Marked DÎfïerential version of CAH ( Eckrnan, 1977) was apparent
on
specific task wmponents involving disparate betwen-language
features
-
such as consonant clusters and specific sounds. For instance,
the
performance of the Cantonese as L1 children on the AAS tasu,
h;idving
initial consonant clusters, was significantly inferior at times
1 and 2.
Since there are fewer consonant clusters in their L i , they are
more
marked for the Cantonese children and therefore acquired more
slowly.
At the same tirne, performance of the Cantonese children on the
AN items
on the RD task at Time 1 appeared slightly higher than for the
LI group,
yet not significantly so. Both language groups obtained
generally low
scores on this task, probably due to the difficulty of the
specific phonerne
which is nomally not mastered by children at this age. On the
other
hand, the performance at Times 1 and 2 of the Cantonese children
was
inferior to that of the L I children on a task requiring them to
strip a sound
from a consonant cluster and blending the remaining phonemes to
make
a wrd (e.g., AAS on sets 2, 3 and 4). In other item sets,
however,
differences betwwn English as L I and Cantonese as L1 diildren
did not
favour consistently the group that were perfoming in their L i ,
namely,
English. It appesrs aiat at least in the beginning stages of
leaming to
read, the Li phonology plays a part in hindering the acquisition
of
unfamiliar phonology in the second language. In this case,
initial
consonant clusters are more marked features for Cantonese as L
I
-
children, and they are more difficult to acquire. Hence, in
agreement with
the Contrastive Analysis Hypathesis and Marked Differential
HypiShesis,
the Cantonese children showd interference from their native
language
specifically in their difFïculty in leaming the specific
phonemes. These
results replicate the findings of some cross-language studies
(e.g., Bruck
& Genesee, 1993; Fashola, 1994) which showd that L l
interference
disappeared in the L2 children with instruction and exposure to
the L2
over time. By Time 3, one year elapsed and Cantonese
children
perfomed at least as well as the L i children on the Auditory
Analytic
Ski Ils, Auditory Discrimination and Rhyme Discrimination
tasks.
Support for the hypothesis of positive transfer from L I to L2
for
language specific features as proposed by the Marked
Differential
Hypothesis of CAH (Eckman, 1977) and Feature Cornpetition
and
Segment Transfer (FCM) (Hancin-Bhatt, 1994) was demonstrated on
two
phonological masures (Oddity and Rhyme Discrimination).
Robust
evidence was found in the significantly better scores of the L2
group
across al1 three urnes, particularly on the latter task. As
discussed earlier,
the presence of an L i factor was apparent, which in this case
probably
accounts for the better performance by the L2 children. The
monosyllabic
and tonal nature of Cantonese probably makes rhymes such as lapl
more
-
prominent and less marked for the Cantonese group. As was
proposed in
the hypothesis of Marked Differential of CAH (Eckrnan, 1977) and
FCM
(Hancin-Bhatt, 1994), less marked features facilitate positive
transfer.
Hence, in this case, the native language possibly leads to a
positive
transfer instantiated by the Cantonese children in that they are
more
sensitive to the rhyme in single syllable wrds used on this
task. What is
interesting ta note is that despite the fact that the lshl
phoneme is absent
in their native language, Cantonese diildren perfomed better
in
discriminating the rhyme leshl than the L I group at Times 1 and
2 (F (1,
57)= 72.47, pc.001). The discrimination of the /SN phoneme,
however,
was not significantly different in either language group on the
Oddity task,
which involved the phoneme in the initial position. It is
possible that for
both language groups, a phoneme in the final position of a w r d
is more
salient and easily distinguishable than the same phoneme in the
initial
position (Segal-Seiden, 1997). According to the FCM, a specific
feature
higher in the hierachial structure in the phonology is likely to
be more
salient than another feature that is in a lower position. A
superior
performance of the Cantonese diildren in discriminating a
'strange"
phoneme in the final position provides another piece of evidence
for
positive transfer. When the tom and the syllable are more
salient
-
features than a phoneme in the native language, it likely leads
to positive
leaming effects by facilitating better discrimination for the
rhyme.
Moderate support to Hypothesis 4, that positive transfer would
ocair
in similar betwsen-language features as hypothesized by the
Marked
Differential Hypothesis of CAH (Eckman, 1977) and FCM
(Hancin-Bhatt,
1994) was obtained from the present study. On the other hand,
there was
no evidence frorn the present study to suggest a simulation
error due to
L I influence on sirnilar between-language features as
demonstrated in
the study by Oller and Ziahosseiny (1 970). The L2 children
appeared to
do better on tasks involving similar between-language features.
For
example, on the first set of items on the Rhyme Discrimination
task, which
contained similar betvveen language phonological features (e.g .
the rhyrne
lapl), Cantonese children scored significantly higher than the L
I group at
Times 1 and 3, as indicated by a significant interaction between
time and
language groups. This result suggests that the Cantonese
children were
probably more sensitive in diswiminating rhymes that exist in
both
languages. Although the Cantonese children appeared superior to
their
L I peers on the cornparison items on both Oddity and Rhyme
Discrimination, the overall differences (al l tasks combined)
wsre not
statistically significant. In other wrds, the fact that
awareness of onset
-
consonants is one skill acquired early in the process of
reading
acquisition (Adams, 1 990, Goswami & Bryant, 1 990, Cha!!, 1
990, Enhri,
1991 ) appears to lie valid for both the Cantonese ESL students
as well as
their L I counterparts.
In sum, this study found that both the L i and L2 children
made
significant gains in their phonological processing skills over
time.
However, compared to their L I counterparts, the
developmental
trajectofies in phonological awareness for the L2 group were
significantly
disparate. L2 children were infefior to their English as LI
counterparts on
phonologically more challenged tasks ai the beginning stages of
the
study. However, despite the fact that their knowledge of the L2
rernained
significantly lower, L2 children were capable of attaining
growth by Time 3
to perfom at least as wsll as, and at times exceeded those for
their
English as L I peers. Results of the present study suggest L i
can cause
interference on the acquisition of disparate phonological
features in L2,
as instantiated in tasks involving consonant blends. At the same
time,
there is also evidence to suggest positive transfer in leaming
of
phonologically more salient features in their L i (Le., rhyme),
although
positive transfer for similar between-language features wsre not
evident.
The present study also confirmed that phonological processing
skills
-
accounted for significant variance for reading measures, even
when the
variance for intelligence is controlled for. On the other hand,
there is also
evidence that non-verbal intelligence affected the leaming of 12
word
recognition skills.
Limitations
Sample size and duration of the present study may lirnit the
generalizability of its findings. As well, there was no atternpt
to explore
the contribution of factors such as the social-econornic status
of the
participants, the impact of instructional practices in different
classrooms,
the extent of previous exposure to the English language and
the
contribution of oral proficiency in English. As a result,
effects, if any, from
these individual and environmental factors on the leaming of
English as a
second language cannot be ascertained. Due to the large number
of
phonalogically disparate between-language features, only a
limited
phonemes wre targeted in the phonological tasks in the present
study.
A more thorough investigation comparing other
behveen-language
features may present a more comprehensive developmental pattern
of the
leaming of English as a second language by Cantonese
children.
-
Relations of Phonological Awareness to Reading and Its
Imp!ications
Results frcm the present study suggested that phonological
features
that are common to tw languages facilitate transfer fraz L I to
L2.
Common between-language phonemes such as /ml and lbl; and
rhymes
such as lapl, /hg/ cm therefore be utilized for assessrnent
of
phonological awareness for children in both L I and L2 language
groups.
Furthemore, it was clear that given sufficient reading
instruction and
exposure, an inferior knowledge of L2 vocabulary did not
significantly
irnpede the capacity for growth in awareness of L2 phonology in
L2
children. Performance on phonological awareness tasks of L2
children at
Grade 1 significantly predicted reading abilities at Grade 2, in
almost
similar magnitude to that of the L I group. Henœ, it would be
valid to
conclude that for L2, as wll as for L I children, difficulties
in manipulating
phonological features in the language as early at Grade 1 can be
used as
an index of at-risk status for having reading difficulties.
The importance of phonological awareness for w r d reading
was
confined in the present study. At least at the initial stages of
leaming to
read, phonological awareness accounted for up to 67% of the
variance on
measures of w r d reading in the L I group, and up to 34% for
the L2
group, even when the variance due to non-verbal intelligence
was
-
controllad for. As was postulated by researchen in the field, in
the
beginning stages of leaming to read, being aware of phonemes in
mrds
is one of the skills facilitating word reading (Adams, 1990;
Chall, 1983,
1996; Ehri, 1995; Gsowami & Bryant, 1990; Huang &
Hinley, 1994;
Stanovich, 1 992; Swirling & Sternberg, 1 994; Wagner, et.
al, 1 993). In
the beginning stage of the present study, Cantonese children
appeared
not to have fully developed knowledge of English phonology and,
in fact,
they were inferior to their L I peers in the knowledge of
English
vocabulary throughout Times 1, 2 and 3. Limited proficiency in
English
probably contributed to deficient knowledge of its phonology,
and
wnsequently, phonological awareness did not account for as
much
variance of reading in the L2 group as it did for native
English-speakers.
An implication of this finding is that both exposure and
instruction are of
equal importance. The present study suggests that in leaming
English as
a second language, increased exposure to the language as well as
to
tasks facilitating phonolog ical processing skills are both
important for
Cantonese ESL students. Phonological awreness is essential
for
leaming to read in a second language as much as it is for the LI
.
Furthemore, second language leamers need additional exposure to
the
language to facilitate their awareness of the phonology. The
present
-
study suggests that teaching strategies and programming for
ESL
students should include both cornponents to ensure their success
in
reading and writing.
-
Table 1. Mean Scores of Phonological Tasks by Time and
Language
Range Time 1 Tirne 2 Time 3
Oddity L1
L2
RD LI
L2
AAS LI
L2
AD LI
L2
Notes: RD, Rhyrne Discrimination Task; AD, Auditory
Discrimination Task; AAS: Auditory Anaiytic Skills Task. LI, Native
Englishspeaking children; L2 Cantonese-speaking ES1 children
Standard Deviation in brackets. Range is measured as the mean
magnitude between minimum and maximum scores.
-
Table 2. Correlations among Phonological Tasks-Combined and
Individual Groups, (Tirne 1 )
N= 33 (Li ); 33 (L2) Oddity RD AD) AAS
Odd ity L I +L2 1 .O0 11 1 .O0 L2 1 .O0
AAS
Note: 1. RD, Rhyme Discrimination Task, AD, Auditory
Discrimination Task,
AAS, Auditory Analytic Skills Task- 2. *p< -05, '* pg.01
-
Table 3. Tne Effects of Time and Group on Phonological TasKs:
MANOVA Summary Table
VVithin Within B e Tirne* Tirne* Subjed Subjed tween Group Group
Multi- Uni- Subj. Multi- Uni- variant variant F variant variant F
value F value value F value F value
Oddity Time 101.12* T l 7 2 147.99- 0.46
RD
AD
AAS
T2T3 43.01-
Time 20.12"" T l T 2 40.98- 2.27
T2T3 0.92
Time 10.57"" T l 7 2 21 .SOM 2.77
T2T3 2.12
Time 18.78- T l T 2 38.13"" O. 50
T2T3 3.23
Note: 1. RD, Rhyme Discrimination Task; AD, Auditory
Discrimination Task,
A M , Auditory Analytic Skills Task. 2. *pc0.05, * p
-
50 Table 4. Mean Percentage Scores of Auditory Analytic Skills
Task by
Se:$
Time 1 Tirne 2 Time 3
Set 1 LI 96.21(9.1) 98.50(6.1) 100 (0) (Syllables) L2 96.3(11.4)
100 (O) 99.1(4.8)
Set 2 LI 6 .91 (44.6) 71.72(33.5) 74.74(37.3) ( Initial L2
34.34(38.6) 48.15(44.7) 70.37(40.7) Consonants)
Set 3 L I 46.46 (42.4) 64.65 (36.3) 60.60 (38.6) (Final L2 53.09
(42.6) 40.74 (37.4) 65.43 (37.5) Consonants)
Set 4 LI 27.27(29.2) 38.38(33.5) 34.34(34.9) (F irst 12 23.45
(30.4) 25.93 (33.8) 46.91 (39.5) Consonant,- initial blend)
Set 5 L I 7.07 (33.4) 14.14 (30.1) 16.16 (32.4) (First L2
22.22(29.2) 24.69(34.1) 41.98(41.9) Consonant, Final Blend)
Set 6 LI 9.09 (25.4) 28.28 (47.2) 19.70 (31.1) (Second L2
22.22(39.2) 22.22(41.3) 42.59(46.9) Consonant- Initial Blend) Note:
Standard Deviation in brackets.
-
Table 5. The Effect of Time and Group on Sets of Auditory
Analytical Skills Task: MANOVA ~ummay Table
Within Within Between
Set 1
Set 2
Set 3
Set 4
Set 5
Set 6
Time
TimeTSL
Time
TirneaESL
Time
TirneeESL
Tirne
Tirne*ESL
Tirne
TirneTSL
Time
Tirne*ESL
Su bject Subject Su bject Multi- Univariant Multi- Variant F
Value Variant F Value F Value
Time 112 8.6" Time 33 Time 1 13 Time 2/3
Time 112 Time U3 Time 1 12 Time 213
Time 1 12 Time 213 Time 1 12 Time Z3
Time 1 /2 Time 2/3 Time 1 12 Time 2/3
Time 1 12 Time U3 Time 1 12 Tme 213
Time 11 2 Time 2/3 Time 1 12 Time 213
-- -
Note: pc .05, "p< .O1, "pc .O01
-
Table 6. Mean scores of Rhyme Discrimination Task by Sets, L I
and L2 -- - -
Time 1 Time 2 Time 3
Setl(lap/) LI 3.16 (1.5) 3.64(1.4) 3.69 (1.4)
L2 3.48 (1.6) 3.33 (1.4) 4.26 (1.2)
Set 2 (Ishl) LI 3.09 (1.4) 4.13 (1.6) 4.47 (1.6)
L2 3.15(1.8) 4.58(1.4) 5.38 (1.2)
set 3 (/th/) 2.28 (1.1 ) 2.47 (1.4) 2.69 (1.4)
L2 2.1 1 (1.4) 2.74 (1.5) 2.93 (1.4)
Note : 1. Set 1, lapl, control items; Set 2, Ishl, target items
with phoneme Ishl;
Set 3, /th/, target items with phoneme /W. 2. Standard Deviation
in brackets.
-
Table 7. The Effect of Time and Group on sets of Rhyme
Discrimination: MANOVA Summary Table
Within Within Between Su bject Subject Subject multi-
uni-variant multi- variant F Value variant F Value F Value
Set1 (lapl) Time 5.52- Time 1 /2 8.1" 0.64 Time 213 0.96
Time*ESL 2.94 Time 1 12 0.29 Time W3 4.70*
Set2 (/sh/) Time 37.47- Tirne112 72.47"" 2.36 Tirne U3 3.30
TirneaESL 2.00 Time 1 /2 4.08" Time 3 3 0.01
Set3 (/th/) Time 3.00 Timell2 5.99" 0.30 Time U3 0.22
TimeeESL 0.45 Time 1 12 0.67 Time U3 0.29
Note: 1. Set 1, lapl, control items; Set 2. Ishî, target items
with phoneme khi;
Set 3, Ahl, target items with phoneme /th/. 2. *p< .05, *
p< -01, - p~ .O01
-
Table 8. Mean Scores of Word Reading Tasks by Time and Language
Groups.
Range Tirne 1 Time 2 Time 3
Word LI 35 5.76(7.27) 11.64(9.22) 14.85(9.92) Attack
L2 38 7.43 (9.27) 12.29 (1 0.23) 19.1 4 (1 3.1)
WRAT- LI 19 4.39(3.94) 7.79(4.61) 9.76(5.36)
L2 18 7.04 (4.88) 9.89 (5.26) 13.30 (5.51)
EXR LI 16 4.55 (4.47) 10.45 (4.98) 1 1.58 (4.52)
L2 16 7.25(6.86) 10.19(5.32) 12.44(4.15)
-
Notes: W h Word Attack, Woodcock Reading Mastery Test, Revised;
WRAT, Wide Range Achievement Test, Revised, Word Reading subtest;
EXR, Experimental Word Reading Task LI, English as First Language ,
L2 , Cantonese group. Range is measured as the mean magnitude
between maximum and minimum scores. Standard Deviation in
brackets
-
Table 9. Conelation among Phonological and Reading Tasks (Time
2)
WA WRAT EXR ODD RD AD AAS
WRAT Ll+K LI L2
EXR Ll+U LI L2
Note: 1. * p
-
56
Table 10. Partial Correlation of Combined Mean z-scores of
Phonological and WA and WRAT Tasks, Contro!!ing for Mean MAT
Scores
r R square r R square
Time 2
O. 72- O.. 52 O. .52+* 0.26
Note: * p< -05, " p< .01, " p< ,001
Table 11. Correlations of Combined z-scores of Phonological and
WA and WRAT Tasks.
Time 1
Time 2
r R square r R square
0.72- 0.52 0.66- 0.43
Time 3 0.84- O, 70 0.65- 0.43
Note: - p4.001, p
-
Figure 1. Mean scores of phonological tasks Note: OD, Oddity , -
1, -2, -3, Times 1, 2, 3; RH, Rhyme Discrimination , AA, Auditory
Analytic Skills, AD, Auditory Discrimination
-
Figure 2. Cornparison of three phonological tasks, Rhyme
Discrimination, Auditory Discrimination and Auditory Analytic
Skills.
Note: RD, Rhyme Discrimination, AD, Auditory Discrimination,
AAS, Auditory Analytic Skills.
-
60 REFERENCES
Adams, M. J. , (1 990). Beginning to Read: Thinking and Leaming
about
p M . Cambridge, Mass.: The MIT Press.
Bnick, M., & Genesee, F., (1995). Phonological awareness in
young
second language leamers. Journal of Child Language, 22,
307-324.
Caravolas, M. & Bnick, M., (1 993). The effect of oral and
witten
language input on children's phonological awareness: A
cross-
language study. Journal of Experimental Child Psychology, 55, 1
- 30.
Chall, J. S., (1983, 1996). Stages ofmding development, Pp.
1-109.
New York: McGraw-Hill..
Cisero, C. A., & Royer, J. M., (1 995). The development and
cross-
language transfer of phonological awareness. Contemporary
Educational Psychology, 20,275-303.
Cummins, J., (1 991 ). Interdependence of first- and
second-language
proficiency in bilingual children. In E. Bialystok, (Ed-),
Language
Prucessing in BiIingual ChiIdren, P p. 70-89. Cam bridge, U. K
:
Cambridge University Press
Cutler, A, Mehler, J e t Noms, D., 8 Seigui, J., (1 992). The
monolingual
-
6 1 nature of speech segmentation by bilinguals. Cognitive
Psychology,
24, 31 8-41 O.
Durgunoglu, A. Y., Nagy, W. E., & Hancin, B. J., (1991).
Cross-language
transfer of phonemic awareness. Journal of EducahOnal
Psychology, 85,453465.
Eckman, F. R., (1 977). Markedness and the contrastive
analysis
hypotheses. Language Leaming, 27, (2), 31 5-330.
Eckrnan, F. R., & Iverson, G. K, (1 993). Sonority and
markedness among
onset clusters in the interlanguage of ESL leamers. Second
Language Research, 9, (3), 234-252.
Ehri, L. C., (1991). Rewnceptualizing the development of sight
word
reading and its relationship to rewding, Pp. 107-1 41.. In A
Fowier,
(Ed.), Development of Speech P m s s & A wareness.
Depreau,
Genany: Mahler.
Fashola, O. S., (1 997). An explanation of the difference betwen
native
English speakers and native Spanish speakers' spelling emrs.
-
62 Paper presented at AERA Conferenœ, 1997.
Geva, E., (1997). Issues in the development of second language
reading:
Implications for instruction and assessment. Paper presented at
the
international Workshop on lntegrating Literacy, Research and
Pradice, March 1 3-1 4, London, England.
Goswami, W., & Bryant, P., (1 990). Phonological Skilis and
Learning to
Read. East Sussex, U. K.: Erlbaurn.
Hancin-Bhatt, B., (1 994). Segment transfer: A consequeme of a
dynamic
system. Second Language Research, 10, (3)) 241 -269.
Hu, C. F. & Catts, H. W., (1993). Phonological rewding as a
universal
process? Reading and WMng: An lnterdisciplinary Journal, 5,
325-
337.
Huang, H. S., & Hanley, J. R, (1994). Phonological awareness
and visual
skills in leaming to read Chinese and English. Cognition, 54,
73-98.
Perfetti, C. A., (1 992). The representation problem in reading
acquisition.
In P. B. Gaugh, L. C., Ehri, & R Trieman (Eds.), Reading
Acquisiaon. New Jersey: Hillsdale. Pp. 1 45-1 73.
Oller, J. W., 8 Ziahosseiny, S. M., (1 990). The contrastive
analysis
hypothesis and spell ing errors. Language Leaming, 20, (2), 1
83-
-
Segal-Seiden, L., (1997). Perception and spelling of strange
speech
sounds by Polish-Canadian L2 speakers of English.
Unpublished
M. A. thesis, Ontario lnstitute of Studies of Education,
University of
Toronto,
Snow. C. E., (1 992). Language proficiency: Towards a
definition. In G.
Appel & H. W., Dechert (Eds. ). A Case Tor PsychololinguisbC
Cases.
Amsterdam, Holland: John Benjamin.
So., Dl (1 989). Learning to read Chinese: Semantic, syntactic,
and
phonological processing skills in nomally achieving and
reading
disabled Chinese children. Unpublished M. A Thesis, Ontario
institute of Studies in Education, University of Toronto.
So, L. K., 8 Dodd, B. J., (1 995). The acquisition of phonology
by
Cantonese-speaking children. Journal of Child Language, 22,
473-
495.
Stanovich, K E., (1 992). Speculations on the causes and
consequences
of individual differences in early reading acquisition. In P. B.
Gough,
L. C., Ehri, Y. R Trieman, (Eds.), Reading acquisition, New
Jersey:
M. Hillsdale.
-
94 Wagner, R. K., Torgesen, J. K, Laughon, P., Simmons, K, 8
Rashotte, C.
A., (1 993). Developrnent of young readers' phonological
pmcessing
abilities. Journal of Educational Psycholagy, 85, (1), 83-1
03.
Wagner, R. K, (1 986). Phonological processing abilities and
reading:
Implications for disabled readen. Journal of Learning
Disebilites,
19, (IO), 623630.
Verhoeven, L. T., (1994). Transfer in bilingual development:
The
linguistic interdependence hypothesis revisited. Language
Learning, 44, (3), 381 -41 5.
Watson, l., (1991). Phonological processing in two languages. In
E.
Bial ysto k ( Ed. ), Language Pmcessing in Bilingual
Chitdmn.
Cambridge: Cambridge University Press.
Whitman, R. L., (1 970). Contrastive analysis: Problems and
procedures.
Language Learning, 20, (2). 1 9 1 -1 97.
-
PON
NES
SHAT
SEN
TAS
THOFE
Z OlJ
COS
SHEM
MAK
THOJE
SHUK
VOLE
SHON
KE7
FIN
THOME
Oddity Task Items
BOK
NEN
SHAM
SHEP
FAD
ZOKE
THOJE
HON
SEP
MAT
THOWG
FUP
THONE
SHONG
FEP
SHlNG
VOAZE
POM
MENG
SAN
SHENG
TAT
VOZ E
THONE
HOD
SHED
NAD
ZOVE
SHUNG
TWOTE
FON
FED
SHlM
THOVE
-
Rhyme Discrimination Task Items
Stimulus Item Target Test
Items
YAM TAN DAT KAM
LIN FI M MIN NIG
POK YOK MOT NOP
NEP KET MEK YEP
VAT NAD TAT UAK
FAP HAP KANG YAN
TESH TES PEF YESH
NEF FESH KES
SESH MES NEF
WEF NES MESH
HES PESH YEF
NESH KES HEF
SOTHE GOVE BOZE POTHE
HOVE THOTHE COZE
POAVE FOVE COTHE
NOVE DOTHE YOZE
NOTHE POVE LOAZ
-
Auditory Discrimination Task Items
thone-von
shen-sen
sen-fen
tow-togg tep-tet
nush-nush
keat he-keev
tekk-tskk
noff-noss
poth-poth
konn-komm
meathe--meez
monn-monn
OZ-402
j ~ o f ~ o o f
bish-biss
lenn--lenn
tom-tonn
t hg-tig
thop-zop
rnak-mag
jekk-jbekk
noz-nov
noove-noove
nesh-neff
zeem-zeem
l up lu t
theak-theak
tas-tas
foom-shoom
ta-tass
zam--vam
theak-theak
nim-nin
-
Auditory Analytic Skills Task Items
Correct Item Question Response
1. Say sunshine Now say it again but, don't say Sun shine. -
2. Say baseball Now Say it again but, don't Say Bal1 base.
3. Say seesaw Now say it again but, don't Say see. Saw 4. Say
picnic Now say it again but, don't say nic. Pic
5. Say leg Now Say it again but, don't say IL/. Egg 6. Say meat
Now say it again but, don't Say /MI. Eat 7. Say hand Now say it
again but, don't say /Hl. And
8. Say pain Now Say it again but, don't say MI. Pay 9. Say keep
Now say it again but, don't sa