You are on page 1of 21

This article was downloaded by: [FU Berlin]

On: 08 May 2015, At: 15:40


Publisher: Routledge
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered
office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

The Language Learning Journal


Publication details, including instructions for authors and
subscription information:
http://www.tandfonline.com/loi/rllj20

The influence of prosodic input in the


second language classroom: does it
stimulate child acquisition of word
order and function words?
a b
Dorota E. Campfield & Victoria A. Murphy
a
Educational Research Institute , Warsaw , Poland
b
Department of Education , University of Oxford , Oxford , UK
Published online: 01 Jul 2013.

Click for updates

To cite this article: Dorota E. Campfield & Victoria A. Murphy (2013): The influence of prosodic
input in the second language classroom: does it stimulate child acquisition of word order and
function words?, The Language Learning Journal, DOI: 10.1080/09571736.2013.807864

To link to this article: http://dx.doi.org/10.1080/09571736.2013.807864

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the
“Content”) contained in the publications on our platform. However, Taylor & Francis,
our agents, and our licensors make no representations or warranties whatsoever as to
the accuracy, completeness, or suitability for any purpose of the Content. Any opinions
and views expressed in this publication are the opinions and views of the authors,
and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content
should not be relied upon and should be independently verified with primary sources
of information. Taylor and Francis shall not be liable for any losses, actions, claims,
proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or
howsoever caused arising directly or indirectly in connection with, in relation to or arising
out of the use of the Content.

This article may be used for research, teaching, and private study purposes. Any
substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing,
systematic supply, or distribution in any form to anyone is expressly forbidden. Terms &
Conditions of access and use can be found at http://www.tandfonline.com/page/terms-
and-conditions
Downloaded by [FU Berlin] at 15:40 08 May 2015
The Language Learning Journal, 2013
http://dx.doi.org/10.1080/09571736.2013.807864

The influence of prosodic input in the second language classroom:


does it stimulate child acquisition of word order and function words?
Dorota E. Campfielda* and Victoria A. Murphyb
a
Educational Research Institute, Warsaw, Poland; bDepartment of Education, University of Oxford,
Oxford, UK

This paper reports on an intervention study with young Polish beginners (mean age: 8
years, 3 months) learning English at school. It seeks to identify whether exposure to
rhythmic input improves knowledge of word order and function words. The ‘prosodic
bootstrapping hypothesis’, relevant in developmental psycholinguistics, provided the
Downloaded by [FU Berlin] at 15:40 08 May 2015

theoretical framework for the study. Eighty-seven children were randomly assigned to
a treatment group exposed to rhythm-salient input in the form of nursery rhymes, a
comparison group exposed to prose input, or a control group with no extra input.
Results established that prosody can be an important factor in second language
acquisition, as in first language acquisition. Children in the treatment group showed
improvement in metalinguistic knowledge of English word order but not of function
words. This has implications for teaching methods and classroom materials.

Introduction
This paper reports on a study of the role of prosody in child second language (L2) acqui-
sition. The aim was to examine whether child beginners learning L2 English in a Polish
instructional setting attend to the prosody of the target language in ways that might help
them ‘bootstrap’ into the grammar of L2. Investigating the role of attending to prosodic
cues in child L2 acquisition is timely, given the increase in the number of children learning
a foreign language in primary classrooms in Europe, together with a reduction in the age at
which they start L2 learning (Cameron 2003). L2 researchers have previously emphasised
the importance of L2 prosody for learners (Brown 1990; Doughty 2005; Vanderplank
1993). A recent study using an elicited imitation task with a similar age group demonstrated
a link between exposure to prosodically salient input and the acquisition of structural
knowledge in L2 English (Campfield and Murphy submitted). The use of songs, poems,
rhymes and rhythmic units with young L2 learners, seen as good practice by many teachers,
may well have a scientific basis.
The theoretical framework for this study was provided by the prosodic bootstrapping
hypothesis, which in developmental psycholinguistics aims to explain how children acquir-
ing their first language (L1) derive grammatical knowledge from a speech signal. The L1
acquisition literature attests to the important role of prosody (stress, rhythm and intonation)
in child-directed speech. The specific bootstrapping models of Gleitman and Wanner (1984)

*Corresponding author. Email: d.campfield@ibe.edu.pl

© 2013 Association for Language Learning


2 D.E. Campfield and V.A. Murphy

and Christophe et al. (1997) demonstrate how, by attending to the prosody of continuous
child-directed speech, L1-acquiring children can begin to construct a mental lexicon and
perform syntactic analysis in the absence of full knowledge of lexis. If attending to prosodic
cues is such an important part of the L1 learning process, then one might assume that it
would be similarly useful for L2-acquiring children. Minimally, therefore, the role of
attending to prosodic cues in L2 development is worth empirical investigation.
Of course, the experiences of L1-acquiring children and children learning an L2 in
instructional settings are fundamentally different. The latter not only receive considerably
less exposure to the target language than L1-acquiring children but also the prosodic cor-
relates present in L1 child-directed speech are often absent in classroom input. Anecdotal
evidence from teacher trainers and the most recent classroom observation studies (Enever
2011; Szulc-Kurpaska 2011) suggest that prosodically rich, continuous speech of the type
directed at L1-acquiring infants is rare in the language classroom, where communication is
often carried out predominantly in learners’ L1. Instructed language acquisition may start
from exposure to continuous speech in rare cases when L2 learners are taught by native tea-
chers who do not speak their pupils’ language. However, even this potentially desirable
Downloaded by [FU Berlin] at 15:40 08 May 2015

situation is fraught with its own complexities, especially when the children in a classroom
share the same L1. Well-intentioned teachers, guided by the desire to simplify input for the
learners, often engage in carefully articulated classroom speech that denies learners vital
experience with segmental cues present in native speech, resulting in the fact that learners
‘… do not learn to rely on the structural information given them by the rhythm of speech but
rely instead upon clear and distinct pronunciation of all vowels and consonants’ (Brown
1990: 47).
The school experiences of young language learners, therefore, often contrast with the
recommendations of researchers and child language acquisition specialists that language
instruction for young learners should focus on exposure to more extended stretches of
spoken language, to discourse at sentence and whole-text level. Such a focus is central
to the ‘learning-centred’ approach, which should provide: ‘children with a broad discourse
and lexical syllabus, that then changes focus as they move into later stages’ (Cameron 2001:
xiii). Cameron (2001) advocates the separation of written and oral skills, with the latter
taught through ‘discourse’ and ‘vocabulary’. By ‘discourse’ the author means not only indi-
vidual sentences but also longer stretches of language. The term denotes language in use,
such as songs, rhymes, stories and stretches of talk. By ‘vocabulary’ she means not only
individual lexical items but also chunks of language, such as phrases. In this top-down
approach, learners first learn example phrases as unanalysed chunks. Linguistic patterns
should stimulate development of grammar.

The prosodic bootstrapping hypothesis


The role of prosody in promoting L1 acquisition has received a lot of attention. Gleitman
and Wanner (1984) argue that three factors, salient to L1-acquiring infants, are important in
the process of prosodic bootstrapping from continuous speech. First, the distinction
between stressed and unstressed syllables underlines the distinction between content and
function words. Second, prosodic cues, such as pauses and the lengthening of the last syl-
lable before pauses, break the stream of speech into units that correspond to phrasal cat-
egories. The salience of prosodic cues to infants is supported by Hirsh-Pasek et al.
(1987), Gerken, Jusczyk and Mandel (1994) and Christophe et al. (1994). The innate
knowledge of the actor–action structure of language enables children to establish which
phrase, distinguishable from the utterance by prosody, is a noun phrase and which is a
The Language Learning Journal 3

verb phrase. The third factor, confirmed by Mandel et al. (1996), is children’s sensitivity to
the sequence of stressed syllables within those units, and therefore content words in utter-
ances. Although in continuous speech prosodic and syntactic boundaries may not always
coincide (Fernald and McRoberts 1996; Steedman 1996), studies have demonstrated that
it is precisely the exaggerated prosody of child-directed speech that assists infants in detect-
ing cues to syntactic phrase boundaries (Garnica 1977; Kemler Nelson et al. 1989). A later
model of phonological bootstrapping (Christophe and Dupoux 1996; Christophe et al.
1997) predicts that the perception of cues to prosodic boundaries results in the creation
of a pre-lexical representation of the speech signal followed by the stripping of function
words. Once this is completed, other segmentation mechanisms such as distributional regu-
larity, phonotactics and lexical bias continue the bootstrapping process (see also Jusczyk
2000).

Research focus
Researchers have highlighted the need to study the relationship between L2 prosody and
Downloaded by [FU Berlin] at 15:40 08 May 2015

acquisition. Doughty (2005: 287), for example, identified ‘a clear research priority in
instructed SLA’ by asking whether L2 learners can:

… return to a mode of processing similar to that used during native language acquisition in
which, at least at first, they pay attention to the cues in the input that are most useful in signal-
ling the relevant lexical, phrasal and syntactic boundaries of the L2, and use that information to
narrow the processing problem space such that other cues may be perceived?

The author argued for an important role for prosody in being able to revert L2 processing
from ‘predictive adult comprehension mode’ to ‘a more efficient acquisition mode’
(Doughty 2005: 299). She did not specify whether, and at which particular level in L2
development, ‘a return to a discovery mode of processing’ would be most beneficial to
the learner. However, given the perceptual acuity of young L2 learners, might not the begin-
ning of the learning process be the best time to expose them to prosodically rich input in the
hope that, like L1-acquiring children, they too would notice prosodic cues to the syntactic
organisation of L2 and exploit them in the process of building L2 representations?
With this question in mind, the present study focused on the very beginning of
instructed L2 learning at the start of formal schooling. The two bootstrapping models dis-
cussed above demonstrate how attention to prosodic cues enables L1-acquiring children to
segment continuous speech in a way that helps them discover the lexis and syntax of their
native language. Since the models predict an early acquisition of word order and function
words (based on the salience of the stressed/unstressed distinction and prosody), the present
study focused on the impact of exposure to rhythmically salient input on the development of
these two aspects of L2 grammar.
Specifically, the question under scrutiny was the following: if L2 learners were exposed
to rhythmically salient L2 input would they, like L1 learners, be able to subconsciously
exploit information in the L2 speech signal to acquire knowledge of word order and func-
tion words? This question was underpinned by two assumptions: that in instructional set-
tings, child L2 learners are not normally exposed to such input; and that when they
begin their experience with L2, usually before the age of 8, they retain the ‘perceptual
acuity’ (Doughty 2005: 298) that allows them to detect prosodic cues in the speech
signal.1 It was hypothesised that exposure to a rhythm-salient intervention would direct
learners to the distinction between stressed and unstressed syllables mapped onto content
4 D.E. Campfield and V.A. Murphy

and function words, respectively. This distinction, coupled with the learners’ ability to
notice pauses and pre-pausal syllable lengthening, which would help in the discovery of
phrase boundaries, and the order in which content words occur in sentences, would
enable L2 learners to acquire knowledge of word order and function words.
It was postulated that if two groups of learners were exposed to continuous speech,
which should encourage listeners to attend to phrase boundaries in utterances, but the treat-
ment group received rhythm-salient input characterised by the distinction between stressed
and unstressed material, learners in the treatment group may notice the stressed–unstressed
distinction and may, therefore, be better placed to learn content words and their sequence in
utterances than learners in the comparison group. Consequently, children in this group
should subsequently demonstrate better knowledge of word order than either children in
a comparison group (exposed to a similar intervention but without the rhythmic salience)
or a control group (receiving no treatment). The stressed/unstressed distinction in the
input may also encourage children in the treatment group to notice unstressed syllables
positioned closely to the salient phrase boundaries. This would afford the treatment
group an advantage over the other two groups in the development of their knowledge of
Downloaded by [FU Berlin] at 15:40 08 May 2015

function words.
The tested hypotheses were thus as follows:

1. Children in the treatment group will demonstrate a better knowledge of word order
than children in the comparison and the control groups by being able to discriminate
better between valid and invalid sentence word order.
2. Children in the treatment group will demonstrate a better knowledge of function
words and their position in phrases than children in the other two groups by
being able to recognise a missing, or an additional, function word in sentences.

Since sensitivity to valid and invalid L2 forms, requiring metalinguistic awareness, was the
main focus of the experiment, grammaticality judgement tasks were chosen as the measure-
ment instrument. Grammaticality judgement tasks were also seen as appropriate for measur-
ing approximate and developing L2 knowledge in young learners in their second year of
learning. Learners may be able to judge whether the grammar of L2 allows or disallows
a particular sentence, even though they may not be able to produce that sentence. Pro-
duction tasks require linguistic skills for successful performance.

Methodology
This research used a mixed design with one independent samples variable (Group) tested at
three levels, with participants randomly assigned into one of three groups: rhythm group,
prose group or control group. Children in the experimental groups (rhythm and prose)
received a 12-hour intervention whilst children in the control group did not receive any
input in addition to their scheduled English lessons at school. Children in the rhythm
group were exposed to rhythm-salient input, such as in nursery rhymes, and children in
the prose group were exposed to the normal speech rhythm of prose, such as in a story-
telling. The lexical and grammatical complexity of the input to the two experimental
groups was specially formulated to contain the same vocabulary and structures. The
same teaching procedures were used. The salience of linguistic rhythm was the only differ-
ence between the two interventions. There was also a related samples variable (Test) tested
at two levels (Pre-test and Post-test) to determine whether responses on the grammaticality
judgement tasks had improved following the intervention.
The Language Learning Journal 5

Participants
The participants were 87 Grade 2 Polish children, 57 boys and 30 girls, with a mean age of
8 years, 3 months, recruited from the sampling frame defined as a radius of 50 kilometres,
which was a realistic distance the researcher could cover. Within that radius, directors of
schools with at least three Grade 2 classes (i.e. up to 90 8-year-olds) were approached
and, of those who agreed to take part in the study, one school was randomly selected. Par-
ental consent for participation was granted for all 87 participants, who were randomly
assigned to the rhythm (treatment), prose (comparison) and control groups. The children
had completed one year of the English programme at school; that is, approximately 54
hours. At the time of the study, the children were just beginning their second year at
school and had two 45-minute English lessons per week. Information on mother’s edu-
cation and extra-curricular exposure to English was collected via a questionnaire. One-
way analysis of variance revealed no differences between the groups in terms of age
(F(2,84) = 1.94, p > .05); gender (χ2(2) = 2.75, p > .05); mother’s education (χ2(4) =
7.46, p > .05); or additional exposure to English (χ2(2) = .94, p > .05).
Downloaded by [FU Berlin] at 15:40 08 May 2015

Experimental materials: screening variables


To obtain a measure of group comparability prior to the intervention, and before the chil-
dren were randomly assigned to the three groups, tests of non-verbal IQ and English pro-
ficiency were carried out. Non-verbal IQ was measured by the Polish standardised version
of the Raven’s Colour Progressive Matrices (1991, 3;11–9;11) with the results interpreted
against Polish norms (Szustrowa and Jaworowska 2003). A χ2 analysis revealed no signifi-
cant differences between the rhythm, prose and control groups in respect of their cognitive
abilities (χ2(6) = 4.26, p > .05), with all participants being in the normal range. To obtain
some measure of the participants’ English proficiency, the British Picture Vocabulary Scale
(Dunn, Whetton and Burley 1997) was carried out with data interpreted according to norms
for pupils with English as an additional language.2 The groups were matched on their
knowledge of receptive English vocabulary as measured by the British Picture Vocabulary
Scale (F(2,82) = .798, p > .05).

Teaching interventions
Twenty well-known nursery rhymes were first chosen to constitute the input to the rhythm-
salient group. Their words and phrases formed the basis for a 12-installment story for the
prose group. The intervention lasted 12 hours, delivered every day in sessions of 45 minutes
each, over the period of three weeks. The length of the intervention was based on Harley’s
study of the effects of instruction on gender in French immersion programmes with children
aged between 7 and 8 years (Harley 1998). Harley exposed her participants to 20 minutes of
experimental input per day for five weeks; that is, eight hours of intervention in total.
The entire teaching effort centred on whole-sentence and whole-text levels with empha-
sis on listening and production. In both experimental groups (rhythm-salient and prose), the
presentation of each lesson followed the same pattern. First, the children were asked to
listen to the text three times whilst looking at the text presented on a big screen via a
data projector. Occasionally, this was varied and the children were asked just to listen to
the text. Next, each phrase/sentence appeared on the screen on its own and children were
taught to repeat it. This was done using the technique of back-chaining based on building
each sentence up from the end, starting with the last word, each time adding the previous
6 D.E. Campfield and V.A. Murphy

word/words such that the speed and fluency of repetition was preserved. This required some
practice because most words were new to the children, as was the practice of producing
whole sentences. The text presented on the screen was supported by visually stimulating
material in the form of pictures and illustrations accompanying the text.

Experimental materials: L2 word order


Polish is a free word-order language and the strict word order of English poses a problem
for Polish learners. To measure the acquisition of L2 word order, 70 stimuli sentences were
constructed: 45 declarative sentences comprising 15 for each of the first phrase patterns 1 to
3 below; 15 negative sentences for phrase pattern 4; and 10 questions for phrase pattern 5:

1: NP (single noun) + VP (single verb) + NP: (PP or AP), e.g. A spider climbed up the
clock.
2: NP (AP) + VP (single verb) + NP (PP or AP), e.g. The little piggy stepped in a
Downloaded by [FU Berlin] at 15:40 08 May 2015

puddle.
3: NP (single noun) + ‘ing’ VP + NP (PP or AP), e.g. Jack is sitting on the roof.
4: Negative sentences, e.g. The poor dog didn’t get a bone.
5: Wh-question word + VP (single verb) + NP (N or PP or AP), e.g. What did the old
dog eat?

The 10 Wh-questions were designed as distractor items aimed to serve as a break to any
response pattern. This was to make sure that the participants were paying attention
during the test, as well as giving them a rest from the test items. McDaniel and Smith
Cairns (1998) demonstrated in L1 research using grammaticality judgement tasks with
3–4 year olds that children tire after 10–15 complex sentences. Given the age of the partici-
pants in the present study, their experience with the school setting and the relatively simple
structure of the stimuli, the number of sentence pairs (i.e. 70) was considered appropriate.
It was important to ensure that the children’s performance reflected their developing
knowledge of word order and that any differences in performance were due to linguistic
structure and not the knowledge of lexis. For this reason, the sentences were constructed
using the lexis to which the children had been exposed during the intervention.
However, successful performance did not require the knowledge of every word in the
test sentences.
The 70 stimuli sentences were presented in grammatical/ungrammatical pairs (e.g.
Mary caught a big fish/Mary a big fish caught) with the ungrammatical sentences con-
structed according to the following phrase-order groups:

1. SOV (Subject Object Verb), e.g. An old clock midnight struck.


2. VOS (Verb Object Subject), e.g. Bit Johnny’s little finger the fish.
3. VSO (Verb Subject Object), e.g. Stayed the kings’ horses at home.
4. OVS (Object Verb Subject), e.g. In a corner sat the two mice.
5. OSV (Object Subject Verb), e.g. Into the bucket the fat baker fell.

Although structures OVS and OSV can occur particularly in stories and rhymes, they are
more unusual in conversation. Participants in the rhythm group were exposed to only
two such examples. Some archaic vocabulary was modified to make rhymes more contem-
porary but word order remained authentic.
The Language Learning Journal 7

There were 12 grammatical/ungrammatical pairs in each phrase-order group. Two pres-


entation orders were prepared: in Order 1, one-half of the stimuli had the grammatical sen-
tence first in each pair and one-half had the ungrammatical sentences first. In Order 2, this
presentation was reversed: one-half of the participants were presented with Order 1 and
one-half with Order 2. The practice and test stimuli were recorded by a male native
speaker of British English. During the recording of the stimuli, care was taken not to
give any aural cues for the grammatical sentences in the grammaticality judgement task
or any negative aural cues for the ungrammatical sentences. The aural version of each
stimulus was presented in as natural an intonation as possible.
The approach of presenting grammatical/ungrammatical pairs was adopted in preference
to presenting individual sentences independently of each other for the following reasons:

1. Presenting similar sentences simultaneously might have helped the children to focus
on the differences between them, thus facilitating the use of strategies for making
grammaticality judgements.
2. At such an early stage in L2 acquisition, the participants might have been confused
Downloaded by [FU Berlin] at 15:40 08 May 2015

by being asked to judge a sentence and then, a few items later, being asked to judge a
similar sentence again.
3. There was no need to ‘hide’ from the participants what the goal of the experiment
was – on the contrary, they were aware that it was to inform the researcher
whether a sentence was a ‘good’ English sentence, in that an English speaker
could utter such a sentence, or whether it was a ‘bad’ English sentence (McDaniel
and Smith Cairns 1998).

Experimental materials: function words


Sixty-four stimuli grammatical/ungrammatical sentence pairs were constructed: 32 ungram-
matical sentences had missing function words and 32 had additional function words. One-
half of the sentences with a missing function word had a missing article and one-half a
missing preposition. Similarly, one-half of the sentences with added function word had
an additional article and one-half an additional preposition. Table 1 shows the structure
of the materials and example sentences.3
Four of the sentences with missing articles had articles missing in the Subject position
and the rest in the Object position. It was generally difficult to construct sentences with
articles missing in the Subject position because Mouse, Cat, Dog and Farmer without
articles could potentially refer to characters. Of the 16 sentences with an added article,

Table 1. Dependent variables with examples of grammatical/ungrammatical sentence pairs.


Function-word set Sub-set Example Number of sentences
Articles Added The wolf is wearing grandma’s clothes. 16
The a wolf is wearing grandma’s clothes.
Missing Mary caught a big fish. 16
Mary caught big fish.
Prepositions Added An old spider was sleeping in the sun. 16
An old spider was sleeping in of the sun.
Missing The little piggy stepped in a puddle. 16
The little piggy stepped a puddle.
Note: Total number of sentence pairs: 64.
8 D.E. Campfield and V.A. Murphy

five had an article added in the Subject position and the rest in the Object position. Of the
five sentences with the article added in the Subject position, three were placed after the
existing article (e.g. The mouse screamed for help/The a mouse screamed for help) and
two were placed before the existing article (e.g. A spider climbed up the spout/The a
spider climbed up the spout). Of the remaining 11 sentences with the article in the
Object position, six had the added article placed after the existing one (e.g. The two mice
sat in a corner/The two mice sat in a the corner) and five with the added article before
the existing article (e.g. A spider is eating a strawberry ice-cream/The spider is eating
the a strawberry ice-cream).
Of the 16 sentences with an added preposition, eight had prepositions placed after the
existing preposition (e.g. They ate breakfast under a big tree/They ate breakfast under with
a big tree) and eight had prepositions before the existing preposition (e.g. An old spider
read the book with Mary/An old spider read the book to with Mary). As with the word-
order grammaticality judgement task, the sentences for the function-words task were con-
structed using the lexis that the children had been exposed to during the intervention but
successful performance did not require the knowledge of every word in the test sentences
Downloaded by [FU Berlin] at 15:40 08 May 2015

and some sentences contained words not used during the intervention.
Two presentation orders were prepared: in Order 1, one-half of the stimuli had the gram-
matical sentence first in each pair and one-half had the ungrammatical sentences first. In
Order 2 this presentation was reversed: one-half of the participants were presented with
Order 1 and one-half with Order 2. The stimuli were recorded by a male native speaker
of British English.

Procedure
The procedure was the same for both the word-order and the function-words grammaticality
judgement tasks. The word-order test was administered to all the participants first, followed
by the function-words test. The pre-intervention tests were carried out by one of the
researchers. However, to ensure test reliability, post-intervention testing was carried out
by two psychologists trained in child testing, one of whom administered the word-order
post-test and the other the function-words post-test. Children were tested individually in
their school, in a quiet room designated for the duration of the study. The test was delivered
via a laptop computer with the child sitting in front of the laptop and the tester to the left.
Instructions appeared on the screen in Polish and were simultaneously given orally, also
in Polish. The children were told that they would hear and see on the screen two sentences
simultaneously. One of the two sentences was a ‘good’ English sentence, in the sense that
an English speaker could utter such a sentence, and one was a ‘bad’ English sentence, in the
sense that an English speaker is unlikely to utter it. The participants were then told to press a
key to indicate which of the two sentences was good. This instruction was repeated to make
sure that the child understood how to respond. The children were told that some of the sen-
tence pairs would be questions. This was followed by a practice session consisting of five
pairs of sentences after which the tester asked the child whether the instructions were under-
stood, whether the child was happy to proceed with the task or whether the child had any
questions. The participants were told that if they felt tired at any point and needed a break
they were to press the escape key. Pressing the escape key again resumed the test.
As already mentioned, each pair of sentences appeared on the laptop screen simul-
taneously with the voice file for each pair. The choice of two stimulus presentation modal-
ities was based on findings from other research using grammaticality judgement tasks
(Murphy 1997), which suggests that performance improves if the stimuli are presented in
The Language Learning Journal 9

two modalities (i.e. aural and visual). The children were given as much time as they liked to
respond as this has also been shown to improve performance (Murphy 1997). Pressing
either of the response keys activated the next screen with the next pair of sentences. If a
child did not make a response, the next pair of stimuli was presented after 2000 millise-
conds. The test lasted approximately 25–30 minutes per child, including instructions, the
practice session and a break.

Results
L2 word-order acquisition
There were no significant differences in performance between the groups before the inter-
vention (F(2) = .42, p > .05). Post-intervention analysis was based on correct judgement
data obtained from 81 of the 87 children taking part in the study.4 Table 2 shows group
descriptive statistics for the scores at pre-intervention and post-intervention. The presen-
tation order of ungrammatical sentences was shown not to influence children’s responses
Downloaded by [FU Berlin] at 15:40 08 May 2015

(F(5, 75) = .88, p > .05). Therefore, subsequent analysis was carried out on the combined
data.
Intervention had a significant effect on the number of correct grammaticality judge-
ments, after controlling for the effect of pre-intervention responses (F(2, 77) = 4.02,
p < .05, partial η2 = .10).5 Figure 1 shows the mean number of correct grammaticality jud-
gements, with 95% confidence intervals per group at post-intervention. Planned contrasts
revealed that membership of the rhythm group significantly increased the number of
correct grammaticality judgements compared with control (t(77) = 2.59, p < .05, r = .28)
but not compared with the prose group (t(77) = .35, p > .05, r = .05). Figure 2 represents
graphically the influence of intervention on the number of correct grammaticality judge-
ments for the rhythm group. Analysis by item supported this result.6
Additional analysis was carried out on the dataset reduced by five children who scored
two standard deviations above the mean at pre-intervention (one rhythm group member,
three prose group members and one control group member). This analysis confirmed
significance of the intervention on grammaticality judgements (F(2,75) = 5.64, p < .01,
η2 = .13, r = .36, which is a medium effect size). Additionally, the Bonferroni post hoc pro-
cedure confirmed not only the significantly better performance of the rhythm group com-
pared with control (p < .05) but also compared with the prose group (p < .05). Owing to
misgivings about removing three children from the prose group, yet another analysis was
carried out in which all seven children who scored particularly high at pre-intervention
were omitted from the analysis (two rhythm group members, three prose group members
and two control group members). This analysis confirmed the above results.

Table 2. Pre-intervention and post-intervention group statistics: mean of correct judgements of


grammaticality, standard error of the mean and standard deviations.
Pre-intervention Post-intervention
Group M SD SE M SD SE
Rhythm 37.21 5.34 .99 41.68 5.48 1.04
Prose 37.38 5.86 1.09 38.44 7.12 1.37
Control 36.21 4.54 .84 37.04 6.99 1.37
Note: Maximum score: 60.
10 D.E. Campfield and V.A. Murphy
Downloaded by [FU Berlin] at 15:40 08 May 2015

Figure 1. Bar chart of the group means of correct post-intervention grammaticality judgements
showing 95% confidence intervals. Maximum score: 60.

The structure of ungrammatical sentences had an influence on children’s responses and


there was a significant difference between groups in this respect (F(5, 70)= 3.16, p < .05,
partial η2 = .20).7 Table 3 shows descriptive statistics for the number of correct judgements
in each phrase set.
Looking for the loci of these differences, tests of between-subject effects showed sig-
nificant differences between groups in sets where ungrammatical sentences were of the
OSV structure (F(2,73) = 6.38, p < .01, partial η2 = .15), whilst parameter estimates

Figure 2. Means of correct pre-intervention and post-intervention judgements of grammaticality,


per group. Maximum score: 60.
The Language Learning Journal 11

Table 3. Post-intervention phrase-order set statistics by group: mean of correct judgements of


grammaticality, standard error of the mean and standard deviations.
Group Phrase-order set M SD SE
Rhythm OSV 9.21 1.618 .328
OVS 7.43 1.952 .320
SOV 8.43 2.026 .408
VOS 8.54 2.009 .372
VSO 8.07 2.142 .399
Prose OSV 8.04 1.911 .339
OVS 7.04 1.743 .331
SOV 7.89 1.948 .422
VOS 7.85 2.179 .385
VSO 7.63 2.097 .414
Control OSV 7.46 2.044 .343
OVS 7.38 1.791 .335
SOV 6.96 2.408 .427
VOS 7.31 2.035 .389
VSO 7.92 2.382 .418
Downloaded by [FU Berlin] at 15:40 08 May 2015

Note: Maximum score: 12.

comparing each experimental group with control pointed towards differences in perform-
ance between the rhythm and control group as follows:

1. OSV, t(79) = 3.39, p < .01, r = .36, which is a medium effect size.
2. SOV, t(79) = 2.36, p < .05, r = .26, which approaches a medium effect size.
3. VOS, t(79) = 1.98 at p = .052, which is marginally significant.

This result is supported by an earlier study of word-order acquisition (Campfield 2006),


which demonstrated that the participants’ grammar disallowed sentences with the verb
phrase in the final position.8 Data given in the Appendix demonstrate that placing the
verb in the final, as opposed to the initial, position in ungrammatical sentences evoked sig-
nificantly more correct grammaticality judgements from the rhythm group, compared with
the control group.

L2 acquisition of function words


The sequence of presenting ungrammatical sentences had no influence on children’s
responses, either before (F(2, 84) =.67, p > .05) or after (F(2, 84) = .10, p > .05) the inter-
vention. The analysis was therefore carried out with the combined data. Post-intervention
analysis was based on correct judgement data obtained from 76 of the 87 children taking
part in the study.9 Table 4 shows group statistics for pre-intervention and post-intervention
data and Figure 3 shows group means of correct pre-intervention and post-intervention
grammaticality judgements (for the article and preposition sets combined).
The analysis revealed that there were no significant group differences in the overall
number of correct grammaticality judgements on function words; that is, in the article
and preposition sets combined (F(2, 72) = 1.45, p > .05, partial η2 = .04).10
Separate analysis was carried out on the article and preposition subsets but no signifi-
cant differences were revealed between groups in their responses to either the article set
(F(2, 72) = .39, p > .05, partial η2 = .01) or the preposition set (F(2, 72) =2.17, p > .05,
12 D.E. Campfield and V.A. Murphy

Table 4. Mean of correct pre-intervention and post-intervention judgements of grammaticality and


standard errors of the mean for all function words, per group.
Group M SE SD
Pre-intervention Rhythm 31.76 .91 .87
Prose 32.34 .89 .76
Control 32.93 1.03 .81
Post-intervention Rhythm 34.51 .87 4.5
Prose 34.44 .76 5.4
Control 32.42 .84 4.8
Note: Maximum score: 64.
Downloaded by [FU Berlin] at 15:40 08 May 2015

Figure 3. The mean of correct pre-intervention and post-intervention judgements of grammaticality,


overall (article and preposition sets combined), for each group. Maximum score: 64.

partial η2 = .06).11 This indicates that the intervention did not significantly influence chil-
dren’s performance on assessing the grammaticality of either function-word set. However,
in the preposition set, although the main analysis did not reveal any significant post-inter-
vention group differences, looking at the 95% confidence intervals in Figure 4 for pre-inter-
vention and post-intervention scores for the rhythm group, it is clear that the error bars
overlap only marginally. This indicates some post-intervention improvement for the
rhythm group, which is supported by planned contrasts revealing that membership of the
rhythm group significantly increased the number of correct grammaticality judgements in
the preposition set compared with control (t(72) = 2.04, p < .05, r = .24, approaching a
medium effect size). Prose group membership did not significantly increase the number
of correct judgements compared with control (t(72) = 1.51, p > .05, r = .18).
To summarise, the results indicate that the intervention significantly increased the
number of correct grammaticality judgements with respect to word order. The rhythm
group outperformed the control group in all phrase-order sets. A closer analysis of the
loci and structures of the ungrammatical sentences revealed that the rhythm group outper-
formed the other two groups on the Verb Last structures. The intervention did not signifi-
cantly influence children’s performance on the function-words task.
The Language Learning Journal 13
Downloaded by [FU Berlin] at 15:40 08 May 2015

Figure 4. Error bars showing each group’s 95% confidence intervals for scores of correct judge-
ments of grammaticality in the pre-intervention and post-intervention preposition sets. 1 = rhythm,
2 = prose, 3 = control.

Based on the work of Cairns et al. (2006: 218), the children in the research sample were
beyond the stage of developing ‘psycholinguistic processing operations’ necessary for
metalinguistic skills in L1. All children were in the normal range on the test of non-
verbal IQ measured by the Polish standardised version of the Raven’s Colour Progressive
Matrices (Raven 1991, 3;11–9;11). The one child with learning difficulties was excluded
from the analysis. Grammatical knowledge was not explicitly taught during the intervention
and was not part of the English course at school. Furthermore, spontaneous comments by
the children during the experiment (such as ‘this feels right’, ‘this sentence is good because
it sounds right’, ‘this sentence is correct because all the words rhyme together’, ‘I don’t
know why but this sentence is bad’) confirmed that they were not basing their decisions
on prescribed rules.

Discussion
The interventions employed were not only brief but also did not involve explicit teaching of
vocabulary, English word order or function words. The aim was to discover whether
exposure to rhythm-salient input alone would direct learners to the distinction between
stressed and unstressed syllables, resulting in enhanced knowledge of English word
order and function words. It should be clearly noted that the aspects of L2 grammar
measured were not taught explicitly during the intervention.
Data analysis showed that the rhythm-salient intervention successfully improved
knowledge of word order. This means that the first hypothesis was supported, albeit with
certain qualifications. A particular pattern of ungrammatical sentences (i.e. OSV and
SOV) evoked more correct grammaticality judgements in the rhythm group, suggesting
that the children’s developing L2 grammar disallowed sentences with the verb phrase in
14 D.E. Campfield and V.A. Murphy

the final position in a sentence. Analysis on the full dataset only showed significant differ-
ences between the rhythm and control groups. However, elimination of high scorers pre-
intervention showed a clearer result. Under these conditions the rhythm group also
scored significantly better than prose.
On what basis were the children making judgements on test sentences? Odlin (1990)
suggested that where basic word order is concerned, learners, especially beginners, tend
to transfer their L1 word order. It is likely that the children were using L1 translation to
work out the meaning of sentences and it is difficult to reject some influence of L1 on jud-
gements of whether the sentences were acceptable. However, given that word order in
Polish is free, a non-significant group result might have been expected, particularly after
only a brief intervention. The result of a greater number of correct grammaticality judge-
ments in the OSV and SOV ungrammatical sentences is therefore intriguing. Sentences
with the verb in the final position (SOV) together with the middle position (SVO and
OVS) sound more natural to native-Polish speakers than sentences with the verb in the
initial position (VSO), the latter being reserved mostly for children’s songs, poems and
nursery rhymes. Therefore, an explanation with reference to a possible L1 word-order trans-
Downloaded by [FU Berlin] at 15:40 08 May 2015

fer would not be entirely satisfactory.


It seems more appropriate to look for an explanation to the developing L2 grammar of
children in the treatment group: rejecting the verb-last order may be an early stage in the
process of implicit acquisition of English word order by Polish child beginner learners. It
is possible that a longer rhythm-salient intervention might show significantly more
correct grammaticality judgements, not only with the SOV and OSV order but also in
other ungrammatical order groups. As already mentioned above, children’s spontaneous
comments during the experiment suggest that they were relying on ‘feel’ to make decisions
on the grammaticality of test sentences. We believe this is evidence that their judgements
were based on ‘abstract linguistic competence of some type’ (Goss, Zhang and Lantolf
1994: 276).
The rhythm-salient intervention did not appear to improve knowledge of function
words. However, in the preposition set, results showed that membership of the rhythm
group but not the prose group, increased the number of correct grammaticality judgements
compared with the control group. This difference in the perception of articles versus prepo-
sitions may be due to the fact that Polish has prepositions but no articles. The rhythm-salient
intervention might have instantiated some new knowledge of prepositions and reinforced
knowledge acquired earlier, since learners are exposed to some explicit teaching of
English prepositions in their first year at school. Additionally, prepositions were semanti-
cally meaningful to the participants, whilst in the absence of articles in Polish ‘the’ and
‘a’ were not. However, the prosodic bootstrapping hypothesis predicts that they should
be noticed because of their acoustic similarity with other function words, the prosodic prop-
erties cued into the speech signal in the form of the stressed/non-stressed distinction and
their position at the borders of prosodic units.
How, then, can the results with respect to function words be explained? First, the inter-
vention might have been successful in sensitising the children to the existence of articles but
its duration was insufficient to enable the participants to start forming representations of
these words in the mental lexicon. This could also have been the case with prepositions.
The difference between the rhythm group and the other two groups is suggestive of
some influence of the rhythmic input on the knowledge of prepositions. Second, the inter-
vention might have been successful in inducing some approximate knowledge of function
words, expressed only as the ability to notice them but the grammaticality judgement task
was not able to tap this knowledge. In other words, participants might have been able only
The Language Learning Journal 15

to form pre-lexical representations of function words. In this case, a different task might be
more successful at tapping these pre-lexical representations. A grammaticality judgement
task is post-lexical, so an online task, similar to the phoneme detection task used by Chris-
tophe and Dupoux (1996) to study adult speech processing, might have been a better
alternative.
A study involving children younger than age 8 is called for to assess the influence of
linguistic rhythm on the acquisition of function words. Newport (1982) suggested that
only learners exposed to function words early in childhood demonstrate native-like
fluency whilst children exposed even at the age of 8 show deficiency in their command.
Research into L2 acquisition of articles (Ionin, Zubizarreta and Philippov 2009; Zdorenko
and Paradis 2008) seems to support this claim. Age of first exposure influences accuracy.
An increased age of exposure is associated with reduced accuracy both in production
and perception of function words.
Two alternative explanations for the outcome of the function word task are possible.
The first is that the function word post-test was administered shortly after the word order
post-test and just before Christmas. Children are excited or distracted before Christmas
Downloaded by [FU Berlin] at 15:40 08 May 2015

and this may not have been conducive to good performance. Another explanation may be
related to difficulties experienced by the tester. Conditions for her work were less favour-
able than those of her colleague owing to office availability. Stress may have adversely
influenced rapport with the children and their performance. These two factors may
equally have contributed to the control group’s lower scores on the second round of
testing, visible in Figure 3. Those carrying out the post-intervention testing were psy-
chologists who had been trained in child testing and experienced in working with chil-
dren. The testers had been introduced to the children prior to testing by the researcher. In
future it would be worth considering the researcher with whom children establish the
closest rapport being involved with pastoral care of participants pre-testing and post-
testing because this will inspire greater security and more confidence.

Conclusions
The motivation for the study arose from two sources. The first was the wealth of literature
attesting to the importance of prosody in L1 acquisition and a resulting question of whether
young L2 learners might benefit from prosodically rich input delivered in instructional set-
tings. The second was a renewed debate about how best to teach a foreign language to
young beginners in instructional settings – a debate encouraged by an increase in the
number of children acquiring a foreign language in primary classrooms in Europe and a
reduction in the age at which they start L2 learning.
The interventions followed the model of foreign language learning suggested by
Cameron (2001), which exposes young beginners to stretches of L2, to L2 discourse at
sentence and whole-text level. Prosody plays an important role in this process, stimulating
the reversion of L2 processing from ‘predictive adult comprehension mode’ to ‘a more
efficient acquisition mode’ (Doughty 2005: 299). The intervention materials in the
present study were constructed to attempt to mimic some features of the richness and
complexity of the input that L1-acquiring children receive. It was obvious to the
researcher that the children enjoyed being able to engage with the content of these
materials, without the need to understand everything; they were able to extract
meaning from approximate understanding.
Gleitman and Wanner (1984: 243) have argued that, ‘for the complex organisms
acquiring complex systems and processes … it is the representation system, far more
16 D.E. Campfield and V.A. Murphy

pervasively than the general inductive system, that must bear the major burden for learn-
ing’. They have also argued that ‘complex organisms’ are ‘biased to represent the
speech signal in certain ways’ (1984: 229). Since the present study suggests that
young L2 beginners seem predisposed to benefit from prosodic cues in the speech
signal and may be able to learn some aspects of L2 structure inductively, classroom
instruction should provide the type of input that, to paraphrase Gleitman and Wanner
(1984), bears the major burden for learning and provides the right environment for
the ‘general inductive system’ to function.
The study suggested that from exposure to prosodically rich L2 input (rhymes, stories
and stretches of L2 talk), young children may be able to perceive some patterns and regu-
larities of the target language and engage in rudimentary syntactic analysis, allowing them
to acquire meaning implicitly and demonstrate an improved knowledge of word order and
possibly also of function words. We are confident that the children’s grammaticality judge-
ments were not formed on the basis of prescribed rules or L1 transfer but reflect some kind
of genuine linguistic intuition in L2.
Downloaded by [FU Berlin] at 15:40 08 May 2015

Notes
1. Doughty (2005: 298) claims that even adult learners may retain perceptual acuity to enable them
to hear ‘those elements of surface structure that are so critical to language acquisition’.
2. The Cambridge Young Learners English Starters Test (Cambridge ESOL 2007) was piloted
and rejected because no meaningful data could be obtained since the children found the
test too difficult. While the British Picture Vocabulary Scale is not intended as a test of recep-
tive English vocabulary for children with English as a foreign language, using this measure
did allow for ensuring no differences between the three groups on this important aspect of
L2 knowledge.
3. The full list of stimuli sentences is available on request from the authors.
4. One child in the rhythm-salient group moved school during the intervention. In the prose group,
the post-intervention test was not administered to one child due to insufficient attendance during
intervention and data of another child were lost due to tester error. In the control group, the post-
intervention test was not administered to one child due to sickness and the data of two children
were lost due to tester error.
5. Univariate analysis of covariance with pre-intervention responses as a covariate was employed
to assess the effect of the intervention on children’s performance. The covariate was signifi-
cantly related to post-intervention scores (F(1, 77) = 21.63, p < .01, r = .47).
6. Analysis by item revealed a significant effect of the intervention on post-intervention scores
after controlling for the effect of pre-intervention scores (F(2, 5666) = 8.83, p < .05, partial
η 2 = .003). Planned contrasts revealed that being in the rhythm group significantly increased
the number of correct judgements of grammaticality, after the intervention, in comparison
with being in the control group (t(5664) = 4.05, r = .06).
7. Multivariate analysis of covariance with pre-intervention scores in each phrase set as covariates.
8. When presented with the ungrammatical sentences in the grammatical/ungrammatical pairs with
the verb in the final position (M = 6.71, SE = 0.35), children made more correct grammaticality
judgements than when presented with the verb first (M = 5.87, SE = 0.27), z = –2.067, p < 0.05,
r = –0.26).
9. In the rhythm group, one child moved school during the intervention and one was not available
for post-testing due to sickness. In the prose group, the post-intervention test was not adminis-
tered to one child due to insufficient attendance during intervention. In the control group, the
post-intervention test was not administered to eight children due to tester sickness on the
final two days assigned for testing before the school Christmas break.
10. Univariate analysis of covariance with correct pre-intervention judgements as a covariate.
11. Univariate analysis of covariance with pre-intervention scores in the preposition and the article
sets as covariates.
The Language Learning Journal 17

References
Brown, G. 1990. Listening to Spoken English. London: Longman.
Cairns, H.S., G. Schlisselberg, D. Waltzman and D. McDaniel. 2006. Development of a metalinguistic
skill: judging the grammaticality of sentences. Communication Disorders Quarterly 27, no. 4:
213–20.
Cambridge ESOL. 2007. Cambridge Young Learners English Test Starters 1 Student Book:
Examination Papers from the University of Cambridge ESOL Examination. Cambridge:
Cambridge University Press.
Cameron, L. 2001. Teaching Languages to Young Learners. Cambridge: Cambridge University Press.
Cameron, L. 2003. Challenges for ELT from the expansion in teaching children. ELT Journal 57,
no. 2: 105–12.
Campfield, D.E. 2006. Assessing ESL children’s knowledge of word order. MSc diss., Department of
Education, University of Oxford.
Campfield, D.E. and V.A. Murphy. submitted. Elicited Imitation in Search of the Influence of
Linguistic Rhythm on Child L2.
Christophe, A. and E. Dupoux. 1996. Bootstrapping lexical acquisition: the role of prosodic structure.
The Linguistic Review 13: 383–412.
Christophe, A., E. Dupoux, J. Bertoncini and J. Mehler. 1994. Do infants perceive word boundaries?
An empirical study of the bootstrapping of lexical acquisition. Journal of Acoustical Society of
Downloaded by [FU Berlin] at 15:40 08 May 2015

America 95: 1570–80.


Christophe, A., T. Guasti, M. Nespor, E. Dupoux and B. Van Ooyen. 1997. Reflections on phonolo-
gical bootstrapping: its role for lexical and syntactic acquisition. Language and Cognitive
Processes 12, nos 5/6: 585–612.
Doughty, C.J. 2005. Instructed SLA: constraints, compensation and enhancement. In The Handbook
of Second Language Acquisition, ed. C.J. Doughty and M.H. Long, 256–310. Oxford: Blackwell.
Dunn, L.M., C. Whetton and J. Burley. 1997. British Picture Vocabulary Scale. Windsor, UK: NFER-
Nelson.
Enever, J., ed. 2011. ELLiE. Early Language Learning in Europe. London: British Council.
Fernald, A. and M.G. McRoberts. 1996. Prosodic bootstrapping: a critical analysis of the argument
and the evidence. In Signal to Syntax: Bootstrapping from Speech to Grammar in Early
Acquisition, ed. J.L. Morgan and K. Demuth, 365–88. Mahwah, NJ: Lawrence Erlbaum.
Garnica, O.K. 1977. Some prosodic and paralinguistic features of speech to young children. In Talking
to Children: Language Input and Acquisition, ed. C. Snow and C.A. Ferguson, 63–88.
Cambridge: Cambridge University Press.
Gerken, L., P.W. Jusczyk and D.R. Mandel. 1994. When prosody fails to cue syntactic structure:
9-month-olds’ sensitivity to phonological versus syntactic phrases. Cognition 51: 237–65.
Gleitman, L.R. and E. Wanner. 1984. Richly specified input to language learning. In Adaptive Control
of Ill-defined Systems, ed. O. Selfridge, E.L. Rissland and M. Arbib, 227–49. New York: Plenum
Goss, N., Y. Zhang and J.P. Lantolf. 1994. Two heads may be better than one: activity in second-
language grammaticality judgments. In Research Methodology in Second-language
Acquisition, ed. E.E. Tarone, S.M. Gass and A.D. Cohen, 263–86. Hillsdale, NJ: Lawrence
Erlbaum.
Harley, B. 1998. The role of focus-on-form tasks in promoting child L2 acquisition. In Focus on Form
in Classroom Second Language Acquisition, ed. C.J. Doughty and J.N. Williams, 156–74.
Cambridge: Cambridge University Press.
Hirsh-Pasek, K., D.G.K. Nelson, P.W. Jusczyk, K.W. Cassidy, B. Druss and L. Kennedy. 1987.
Clauses are perceptual units for young infants. Cognition 26: 269–86.
Ionin, T., M.L. Zubizarreta and V. Philippov. 2009. Acquisition of article semantics by child and adult
L2-English learners. Bilingualism: Language and Cognition 12, no. 3: 337–61.
Jusczyk, P. 2000. The Discovery of Spoken Language. Cambridge, MA: MIT Press.
Kemler Nelson, D.G., K. Hirsh-Pasek, P.W. Jusczyk and K. Wright-Cassidy. 1989. How prosodic cues
in motherese might assist language learning. Journal of Child Language 16: 55–68.
Mandel, D.R., D.G. Kemler Nelson and P.W. Jusczyk. 1996. Infants remember the order of words in a
spoken sentence. Cognitive Development 11, no. 2: 181–96.
McDaniel, D. and H. Smith Cairns. 1998. Eliciting judgements of grammaticality and reference. In
Methods for Assessing Children’s Syntax, ed. D. McDaniel, C. McKee and H. Smith Cairns,
233–54. Cambridge, MA: The MIT Press.
18 D.E. Campfield and V.A. Murphy

Murphy, V.A. 1997. The effect of modality on a gramaticality judgement task. Second Language
Research 13, no. 1: 34–65.
Newport, E.L. 1982. Task specificity in language learning? In Language Acquisition: State of the Art,
ed. E. Wanner and L.R. Gleitman, 450–86. New York: Cambridge University Press.
Odlin, T. 1990. Word-order transfer, metalinguistic awareness and constraints on foreign language
learning. In Second Language Acquisition – Foreign Language Learning, ed. B. VanPatten
and J. Lee, 95–117. Clevedon, UK: Multilingual Matters.
Raven, J.C. 1991. Raven’s Colour Progressive Matrices (CPM). Pearson.
Steedman, M. 1996. Phrasal intonation and the acquisition of syntax. In Signal to Syntax:
Bootstrapping from Speech to Grammar in Early Acquisition, ed. J.L. Morgan and K. Demuth,
331–42. Mahwah, NJ: Lawrence Erlbaum.
Szulc-Kurpaska, M. 2011. Raport z badania realizacji podstawy programowej do jezyka obcego przez
dzieci 6-letnie w pierwszej klasie szkoly podstawowej. [Report - Study of the implementation of
the foreign language curriculum for six-year-old children in Grade 1]. Unpublished report, The
Educational Research Institute, Warsaw.
Szustrowa, T. and A. Jaworowska. 2003. Podrecznik do testu matryc Ravena, Pracownia Testow
Psychologicznych [Raven’s Matrices Handbook]. Warsaw: Psychological Test Unit, The Polish
Psychological Association.
Vanderplank, R. 1993. ‘Pacing’ and ‘spacing’ as predictors of difficulty in speaking and understand-
Downloaded by [FU Berlin] at 15:40 08 May 2015

ing English. ELT Journal 4, no. 2: 117–25.


Zdorenko, T. and J. Paradis. 2008. The acquisition of articles in child second language English:
fluctuation, transfer or both? Second Language Research 24, no. 2: 227–50.
The Language Learning Journal 19

Appendix
Table A1 shows group statistics for post-intervention scores collapsed into Verb First (VOS + VSO)
and Verb Last (SOV + OSV) sets. Placing the verb phrase in the final, as opposed to initial, position in
the ungrammatical sentences had an effect on the number of correct grammaticality judgements
(F(2,76) = 6.56, p < .01, partial η2 = .20; multivariate analysis of covariance with pre-intervention
scores in the Verb First and Verb Last sets as covariates). Pre-intervention scores in the Verb First
set were in a relationship of association with post-intervention Verb First scores (r = .33, p < .01)
and Verb Last scores (r = .31, p < .01).

Table A1. Post-intervention Verb First versus Verb Last statistics by group: mean, standard error of
the mean and standard deviation.
Group Verb phrase position M SD SE
Rhythm Verb First 16.61 3.178 .60
Verb Last 17.64 2.896 .58
Prose Verb First 15.48 3.556 .62
Verb Last 15.93 3.452 .60
Downloaded by [FU Berlin] at 15:40 08 May 2015

Control Verb First 15.23 3.658 .63


Verb Last 14.42 3.349 .61
Note: Maximum score: 24.

Tests of between-subjects effects showed that groups differed significantly in the Verb Last set
(F(1, 76) = 6.55, p < .01, partial η2 = .15). Parameter estimates comparing each experimental
group with the control revealed that the rhythm group made significantly more correct grammaticality
judgements in the Verb Last set than the control (t(79) = 3.57, p < .01, r = .37, which is a medium
effect size). The effect of the intervention on the number of correct grammaticality judgements in
the Verb Last set for the rhythm group can also be observed in Figure A1.

Figure A1. Means of correct pre-intervention and post-intervention judgements of grammaticality in


the Verb Last set. Maximum score: 24.

You might also like