Content uploaded by Theodoros Marinis
Author content
All content in this area was uploaded by Theodoros Marinis on Aug 27, 2016
Content may be subject to copyright.
In press: Armon-Lotem, S., de Jomg, J. & Meir, N. (Eds.). Methods for assessing
multilingual children: disentangling bilingualism from Language Impairment. Multilingual
Matters.
Sentence Repetition
Theodoros Marinis1 & Sharon Armon-Lotem2
1University of Reading
2Bar Ilan University
Address for correspondence:
Theodoros Marinis
School of Psychology & Clinical Language Sciences
University of Reading
Reading RG6 6AL
UK
Tel. +44-118-378 7465
e-mail: t.marinis@reading.ac.uk
2
Introduction
Sentence repetition (SRep) tasks have been shown to be very sensitive and specific in
identifying children with language impairment in monolingual populations, among others,
Conti-Ramsden et al. (2001). Sensitivity measures the proportion of children who have
language impairment and score very low in a specific task; specificity, in contrast measures
the proportion of children with typical language development (TLD) who do not score low in
a specific task. Poor sensitivity may lead to under-diagnosis of SLI, whereas poor specificity
may lead to over-diagnosis. In a seminal paper, Conti-Ramsdenet al. (2001) showed that
Sentence Recall from Clinical Evaluation of Language Fundamentals (CELF) 3 (Semel et al.,
1995) had 90% sensitivity and 85% specificity, much higher than measures for non-word
repetition (78% sensitivity, 87% specificity), past tense (74% sensitivity, 89% specificity),
and 3rd singular –s (63% sensitivity, 90% specificity). These figures illustrate nicely that
SRep tasks are more challenging for children with language impairment than other tasks and
this is why 90% of children with language impairment score below the cut-off point. At the
same time, they are more challenging for some children with TLD as well and this is why
15% of monolingual children with TLD scored below the cut-off point, a higher proportion
than on all other tasks. If SRep tasks are challenging for some monolingual children with
TLD, will bilingual children find them even more challenging if they are tested in their
second and non-dominant language?
SRep tasks have been used in bilingual children with variable success when the
language tested was the children’s second and non-dominant language. Some but not all
studies so far have shown effects of language dominance, Age of Onset (AoO), and/or Length
of Exposure (LoE). Verhoeven et al. (2012) report language dominance effects on children`s
SRep performance, while Thordardottir and Brandeker (2013) explore the effect of external
3
factors such as LoE. Armon-Lotem et al. (2011) report that Russian-Hebrew and Russian-
German bilingual groups already performed within the monolingual range on SRep tests in
Hebrew as a second language (L2) and German as a L2 after two years of exposure. Chiat et
al. (2013) report effects of AoA and LoE only for a Russian-Hebrew bilingual group, whereas
for Turkish-English and Russian-German groups, AoA and LoE were not predictive of the
children’s SRep performance. Finally, Gutierrez-Clellen et al. (2006) show that there is no
gap in the global score of Spanish dominant bilingual children with TLD when compared to
monolingual speakers of Spanish.
The variable performance in bilingual children could be caused by a range of factors
related to the populations tested, but also to the tasks used. The vocabulary used in the SRep
tasks may have not been controlled in terms of widely used psycholinguistic factors, such as
frequency, age of acquisition, and imageability; the length of words and sentences may have
been different across studies and the tasks may have included different ranges of structures or
structures that tap into different underlying grammatical processes. Whatever the differences
in findings and possible reasons for these, it is clear that if bilingual children are being tested
in their L2, low performance may reflect low language proficiency because of late AoO and
limited LoE rather than language impairment. To rule this out, it is important to assess both
of the children’s languages, since Specific Language Impairment (SLI) should manifest itself
in all languages spoken by the child. A detailed language history is also necessary in order to
ascertain the AoO and LoE in all languages spoken by the child (see Tuller, this volume).
To date there is a lack of bilingual/multilingual assessments for children in general
and also bilingual/multilingual SRep tasks. COST Action IS0804 addressed this gap by
developing a range of language tasks including SRep tasks across a large number of
languages. The SRep tasks were constructed using a set of principles to provide comparable
results across languages. This allows bilingual children to be assessed in both languages they
4
speak. Difficulties in both languages would strongly indicate language delay or impairment;
typical performance in their dominant language and difficulties in their non-dominant (often
L2) language would indicate low proficiency in that language but no language impairment.
This chapter will start with a section on the rationale of SRep tasks and the domains
they are tapping into (Section 1). This will motivate the principles and designs used for the
SRep tasks developed within the COST action that will be presented in Section 2. Section 3
will present and discuss examples of designs across a number of languages to illustrate how
cross-linguistic differences were taken into account when the parallel versions were created.
Section 4 will present the procedure and scoring schemes used. Section 5 will present pilot
data. The chapter will conclude with Section 6 which includes a discussion of strengths and
limitations of the tasks, future directions and recommendations of aspects that need to be
taken into account when using the SRep tasks with bilingual children for research purposes
and also within a clinical setting.
1. Rationale of SRep tasks
There is a long tradition of using SRep tasks to assess whether or not children or adults have
acquired specific structures in their first language (L1) or their L2, to measure language
proficiency in a foreign language teaching setting (Bley-Vroman & Chaudron, 1994; Jessop,
Suzuki, & Tomita, 2007; Lust, Flynn, & Foley, 1996), and to measure language abilities in a
speech and language therapy setting (Conti-Ramsden et al., 2001) with clinical assessments
of language often including a sentence recall subtest. Imitating someone else’s behavior,
including language, is an innate ability of humans and other species, and repeating sounds,
words, and sentences is often part of uninstructed and instructed teaching and learning
processes. One important issue that has attracted a lot of attention is whether sentence
5
repetition is a passive echoing (parroting) of the stimulus, in which case it would not be
informative about the participant’s language abilities, or whether it is a productive process
that involves the participant’s grammatical system, in which case it would provide a window
into the participant’s language competence. Research from the 70s onwards (Slobin & Welsh,
1973) has addressed this issue and has shown that sentence repetition taps into the learners
implicit knowledge (Ellis, 2005; Erlam, 2006), although test design and in particular sentence
length are crucial in this respect. If sentences are long enough to disallow passive copying,
participants draw on their grammatical system to repeat the sentences; this requires
processing of the incoming stream, analysis and reconstruction of the meaning of the
sentence using their own grammatical and memory systems. As a result, they are not able to
repeat sentences if they have not acquired the specific structures that are being elicited.
However, if the sentences are very short, participants can bypass the decoding/encoding
process and repeat sentences using passive copy, in which case, SRep will measure only the
participants’ memory capacity. This is why SRep tasks usually include sentences that are
relatively long. It has to be noted that length is relative to age, memory capacity, and
language abilities. For example, 5-year-old children with TLD may be able to repeat three-
word utterances of a structure they have not yet fully acquired if they have good memory
skills and the words are familiar to them, so they have strong lexical representations of the
words; this does not mean that children with SLI of the same age will be able to bypass the
decoding/encoding process and repeat sentences using passive copying; their memory
capacity may not be able to support passive copying of a sentence the structure of which has
not been acquired (Fattal, Friedmann, & Fattal-Valevski, 2011) and they may also have
weaker lexical representations. We will now turn to the processes involved when participants
take part in a SRep task that involves relatively long sentences, and thus, taps into the
children’s grammatical system.
6
SRep involves listening to sentences and repeating them verbatim. To perform a SRep
task, participants have to be able to process/analyse the sentence in terms of all levels of
representation (phonological, morpho-syntactic, semantic), extract its meaning and then use
the production system to regenerate the meaning of the sentence from activated
representations in long term memory (Lombardi & Potter, 1992; Potter & Lombardi, 1990;
Potter & Lombardi, 1998). Therefore, accuracy in repeating sentences verbatim depends on
all processes and levels of representation related to comprehension and production and the
ability to store and retrieve language material from memory. Within Baddeley’s (2000)
Multicomponent Working Memory model, in SRep tasks the episodic buffer plays an
important role along with the phonological loop, the central executive and long term memory.
The episodic buffer is a temporary storage system with limited capacity that holds integrated
chunks or episodic representations in a multidimensional code (Baddeley, 2000, 2012). Due
to its limited capacity, the episodic buffer has been suggested to represent a constraint in
language processing (Boyle, Lindell, & Kidd, 2013).
The interaction between different levels of representation and memory has recently
been explored by Polišenská (2011) and Polišenská et al. (under review) using SRep tasks in
English and Czech. The tasks investigated the effect of prosodic structure, semantic
plausibility, syntactic well-formedness, and use of pseudo-sentences (with all lexical items or
only content words/function words replaced by non-words) on children’s ability to repeat
sentences. Polišenská’s study revealed that each one of these factors affects accuracy in SRep
independently, and thus, indicates a close linkage between different levels of representation
and memory. Interestingly, the largest effects were observed in the conditions related to
vocabulary and morpho-syntax. Given that SRep tasks involve language processing at all
levels of representation in both comprehension and production and also several working
memory components, deficits in one or more of those domains and processes may affect
7
performance on this task. Therefore, it is not surprising that children with language
impairment score low in SRep tasks.
2. Principles and design of the LITMUS-SRep tasks
The principles and design of the LITMUS-SRep tasks were based on previous research using
SRep tasks, on the theoretical understanding of the domains that SRep tasks are tapping into,
and also on previous research on the way SLI manifests itself across languages (Leonard,
1998). Cross-linguistic studies have shown that on the one hand, syntactic complexity affects
children with SLI across languages, and as a result, structures that are syntactically complex
are vulnerable cross-linguistically. On the other hand, there are also important cross-linguistic
differences in the way SLI manifests itself across languages, which relate to a large extent to
the morpho-phonological properties of each language. This creates a challenge when trying to
create parallel versions across languages. To address these issues and to establish parallel
versions across languages, we used the following two principles:
1. Include in all SRep tasks a set of syntactically complex structures that have been
shown to be difficult for children with SLI across languages and involve embedding
and/or syntactic movement along with a set of syntactically simple structures as
control structures (language independent structures);
2. Include a set of structures for each language that have been shown to be difficult for
children with SLI in the specific language (language specific structures).
The language independent structures include object wh-questions (Friedmann &
Novogrodsky, 2011) and relative clauses (Adani, van der Lely, Forgiarini, & Guasti, 2010;
8
Friedmann & Novogrodsky, 2004) that have been shown to be challenging in children with
SLI across languages along with conditionals that are not only syntactically but also
semantically complex (Daniel & Klaczynski, 2006). Mono-clausal sentences and bi-clausal
sentences with coordination and subordination were included across languages as control
conditions matched to the syntactically complex sentences in terms of length.
Language specific structures were selected on the basis of previous research showing that
a structure is vulnerable in a specific language in children with SLI, but not in bilingual
children with TLD. For example, for English we selected sentences with auxiliaries and
modals because difficulties with tense marking constitute one of the hallmarks of SLI in
English (Rice & Wexler, 1996), but auxiliaries are a relative strength in bilingual children
with TLD compared to the bound tense marking morphemes –ed and 3rd sg –s (Paradis,
2005). Passives were also selected for English because children with SLI have been found to
have difficulties with passives (Marinis & Saddy, 2013; Marshall, Marinis, & van der Lely,
2007; Montgomery & Evans, 2009; van der Lely, 1996). Passives were also targeted by the
French version in addition to object clitics which are difficult for French speaking children
with SLI but are not expected to be difficult for bilingual children with TLD (see Prévost, this
volume). By contrast, in Hebrew passives are difficult for all speakers being rather infrequent
(Ravid, Landau, & Lovetski, 2003), and in Russian they are particularly difficult for
bilinguals due to difficulties with the case system which is essential for comprehending
passives (Polinsky, 2006). In Hebrew and Russian some aspects of morphosyntax (e.g.,
present tense feminine plural verb morphology in Hebrew or number and gender morphology
in Russian) are central for discriminating children with SLI from children with TLD because
they are sensitive to language impairment but not to bilingualism (Armon-Lotem, 2012);
therefore, they were included in the SRep task developed (see section on the Hebrew and
Russian SRep tasks).
9
In addition to these two principles, we controlled for the length of the sentences and the
vocabulary used to ensure consistency within and between languages and to ensure that the
children’s memory capacity will not be disproportionately affected in particular structures or
languages. The length was controlled in terms of the number of clauses (inclusion of mono-
clausal and bi-clausal sentences), the number of words/morphemes and the number of
syllables in each sentence. In terms of psycholinguistic factors, word frequency, age of
acquisition, and imageability were controlled when such ratings were available in the
languages. Since vocabulary is often a source of difficulty for bilingual children and could
jeopardize their ability to repeat sentences (see Haman et al., this volume), vocabulary was
restricted to frequent early acquired words rather than vocabulary for age-matched
monolingual children.
Finally, several grammatical properties were controlled in each language depending
on the properties of the specific language. For example, in languages with gender marking
(e.g. French and Greek), gender was balanced across the task.
3 . Examples of LITMUS-SRep tasks across languages
LITMUS-SRep tasks have been developed so far in Albanian, Lebanese Arabic, Catalan,
Croatian, Dutch, English, French, German, Standard and Cypriot Greek, Hebrew, Italian,
Lithuanian, Polish, European Portuguese, Russian, Serbian, and Turkish; Danish, Irish,
Malay, Swedish, and Welsh versions are currently in progress.
Sentence repetition tasks were developed in two stages. At the first stage a large
number of sentence types and sentences (approximately 60) were created in each language
based on the principles outlined in Section 2. These were based on the School-Age Sentence
Imitation Test (SASIT) (Marinis, Chiat, Armon-Lotem, Gibbons, & Gipps, 2010) and the
10
Sentence Repetition task by Fattal et al. (2011). The initial SRep tasks were piloted with
monolingual and/or bilingual children with TLD and/or children with SLI. Reliability
analyses for items were used to shorten the tasks and also analyses showing which structures
differentiated maximally between children with TLD and children with SLI. At the second
stage, a smaller number of sentences (approximately 30 in each language) were selected
based on the reliability analyses and evidence of maximum differentiation between children
with TLD and children with SLI. The second stage is ongoing, therefore, this section will
provide examples from the first stage only. We will start with presenting the design of the
English task and compare it with tasks developed in three typologically different languages,
Hebrew, Russian, and French. Each of these languages represents a different language family
with its unique morpho-syntactic properties: rich verbal inflection in Hebrew (a Semitic
language), rich case system in Russian (a Slavic language) and clitics in French (a Romance
language). This variety further allows different degrees of flexibility in word order, while
allowing similar complex structures such as relative clauses.
English SRep task
The English SRep task (Marinis et al., 2010) had 60 sentences targeting 15 structures (4
sentences per structure). The task was organized around 3 blocks with increasing difficulty,
as shown in Table 1.
Table 1: Structures tested in the English SRep task with examples (number of sentences per
structure in brackets)
Block
Structure
Example
1
Subject-verb-object
(SVO) with one
auxiliary or modal (4)
The kitten is chasing the rat up and down.
She can bring the glass to the table.
SVO with negation and
The man wasn't driving the lorry to town.
11
one auxiliary or modal
(4)
He shouldn’t draw the tiger in the tree.
Short actional passives
(4)
She was stopped at the big red lights.
Wh-who/what object
questions (4)
Who did the monkey splash near the water?
What did they find yesterday in the snow?
Bi-clausal sentences:
coordination or
complement clauses
(4)
His sister ran and his father walked.
The cook tried to make the soup in the kitchen.
2
SVO with two
auxiliaries or one
auxiliary and one
modal (4)
The policeman has been looking at us.
She could have waited for them in the street.
SVO with negation,
one auxiliary and one
modal (4)
John won't have talked about it with his father.
They couldn’t have chased the goose by the river.
Long actional and non-
actional passives (4)
The sandwich was eaten by the postman.
She was seen by the doctor in the morning.
wh-object which
questions and oblique
wh-questions (4)
Which picture did he paint at home yesterday?
Who did she give the beautiful rose to?
Bi-clausal sentences:
complement or adjunct
clauses (4)
She thinks that the spider is very small.
The child ate breakfast after he washed his face
3
Object relative clauses
– sentence final (4)
The children enjoyed the sweets that they tasted.
He should wash the baby that the child is patting.
Object relative clause –
centre embedded (4)
The swan that the deer chased knocked over the plant.
The horse that the farmer pushed kicked him in the back.
Conditionals (4)
The people will get a present if they clean the house.
If the kids behave we will go in the garden.
Object clefts with
actives and subject
clefts with passives (4)
It was the boy that the man splashed in the sea.
It was the paper that was damaged by the fire.
Sentences with nouns
taking complements
(4)
Peter had a dream of becoming a bus-driver.
The promise of going to Paris made them happy.
The language independent structures were monoclausal SVO structures (-embedding, -
movement), object wh-questions (-embedding, +movement), bi-clausal sentences with
coordination or subordination (+/-embedding, -movement), object relative clauses
(+embedding, +movement), and conditionals (+embedding, -movement). Object clefts
12
(+embedding, +movement) were included because they involve movement, like object wh-
questions and object relative clauses. This predicted similar difficulties in all three structures.
Subject clefts with passives tested whether or not children with SLI have difficulties with
sentences involving passives across the board. Sentences with nouns taking complements
(+embedding, -movement) tested whether embedding without movement within the noun
phrase (on-a-par with relative clauses) would be challenging for children with SLI. The
language specific structures were the sentences with auxiliaries/modals (with and without
negation) and short and long passives (-embedding, +movement) because as mentioned
earlier, they have been shown to be difficult for monolingual English children with SLI
(Marinis & Saddy, 2013; Marshall et al., 2007; Montgomery & Evans, 2009; Rice & Wexler,
1996; van der Lely, 1996).
To ensure that word frequency and age of acquisition did not act as confounding
factors, the words used in the three blocks were matched in terms of word frequency and age
of acquisition using the CELEX database (Baayen, Piepenbrock, & Gulikers, 1995) and the
MRC Psycholinguistic Database (Fearnley, 1997). The sentences in Blocks 1 and 2 were also
matched in terms of number of words and number of syllables per sentence, but the sentences
in Block 3 were slightly longer in the number of words and syllables than Blocks 1 and 2, to
accommodate their syntactic complexity. Finally, the type of nominal expression was
controlled between the sentence types with the sentences having an equal number of full noun
phrases and pronouns within each sentence type. The sentence types were pseudo-
randomized within each block to ensure that two consecutive test items did not elicit the same
structure. Stimuli sentences varied in length from 7 to 13 syllables and from 7 to 13 words.
Russian SRep task
13
The SRep task in Russian (Meir & Armon-Lotem, 2014) has 56 items of varying complexity
and length. Special attention was given to language specific aspects of verbal and nominal
morphology which are central in Russian and known to be difficult for monolingual children
with SLI. For verb morphology, gender and number were controlled; for nominal
morphology, prepositions and cases were selected to represent a wide range of cases.
The 56 items were chosen from a larger pool of sentences which were tested on
Russian-Hebrew bilingual children with TLD aged from 5;08 to 6;07 (Mean=6;01). All
children grew up speaking Russian at home and Hebrew in the preschool from age 3, and
scored within monolingual norms in at least one of their languages (Armon-Lotem, 2011).
The pilot study showed poor performance of the bilingual children on a number of syntactic
structures (e.g. passives with the agent-phrase, complex sentences with temporal clauses,
oblique relatives, complex sentences with advanced conjunctions). Thus, it was decided to
eliminate these structures as they pose great difficulties to typically developing bilingual
children aged 6. Eleven structures were selected, as shown in Table 2. The structures vary in
terms of embedding and syntactic movement, including sentences without embedding and
movement (simple mono-clausal sentences), sentences with embedding but without
movement (bi-clausal sentences involving co-ordination vs. subordination), sentences with
movement but without embedding (object and oblique which-questions, OVS), and sentences
with embedding and movement (object relative clauses). These structures made the task
comparable with the other languages. Conditionals were added to tap into semantic
knowledge. The structures were grouped by length and complexity into 3 blocks and pseudo-
randomized within each block so that two consecutive test items did not contain the same
target structure or the same lexical items. There were 12 sentences with SVO, the basic word
order in Russian, to enable independent testing of morphosyntax and prepositions and 11
more structures with 4 sentences in each. The same lexical items were deployed across
14
different syntactic structures to minimize the effect of specific lexical items. Stimulus
sentences varied in length from 4 to 9 words.
Table 2: Structures tested in the Russian SRep task with examples (number of sentences per
structure in brackets)
Block
Structure
Example
1
SVO (8)
kotjonok našjol tapočki pod krovat'ju.
kitten.NOM found slippers.ACC under bed.INSTR
‘The kitten found the slipper under the bed.’
SOV (4)
devočka mal'čika udarila v sadu.
girl.NOM boy.ACC hit in gadern.LOC
‘The girl hit the boy in the garden.’
OVS (4)
mamu poceloval syn pered snom.
mother.ACC kissed son.NOM before sleeping.INSTR
‘The son kissed the mother before sleeping.’
2
Biclausal
Sentences with
coordination (4)
tjotja pomyla posudu, a djadja svaril sup.
aunt.NOM washed dishes.ACC and/but uncle.NOM cooked soup.ACC
‘The aunt did the dishes and the uncle made soup.’
Biclausal
Sentences with
subordination (4)
oni uslyšali, čto ja prišjol iz magazina.
they.NOM heard that i.NOM came from shop.GEN
‘They heard that I came-back from the shop.’
Object Questions
(4)
kakuju zebru lošad' dognala v cirke?
which.ACC zebra.ACC horse.NOM caught-up in circus.PREP?
‘Which zebra did the horse catch up with in the circus?’
Oblique Questions
(4)
ot kakoj sobaki ubežala koška?
from which.GEN dog.GEN ran-away cat.NOM
‘From which dog did the cat run-away?’
3
Real Conditionals
(4)
esli ona pročitaet skazku, my pojdjom v park.
if she.NOM will-read story.ACC we.NOM will-go in park.ACC
‘If she reads the fairy-tale, we will go to the park.’
Unreal
Conditionals (4)
my by pošli v kino, esli by sdelali uroki.
we.NOM SUBJ went in cinema if SUBJ did lessions.ACC
‘We would have gone to the cinema, if the brother had done
homework.’
Subject Relatives
(8)
jeto žiraf, kotoryj tolknul krokodila.
this giraffe.NOM which.NOM pushed crocodile.ACC
‘This is a giraffe that pushed the crocodile.’
Object Relatives
(8)
jeto devochka, kotoruju narisovala mama.
this girl.NOM which.ACC drew mother.NOM
‘This is the girl that the mother drew.’
Hebrew SRep task
The Hebrew SRep task (Meir, Walters, & Armon-Lotem, under review) also has 56 items of
varying complexity and length. For language specific morphosyntax, verb morphology
15
included person morphology and plural morphology which were reported to be difficult for
Hebrew speaking children with SLI (Dromi, Leonard, Adam, & Zadunaisky-Ehrlich, 1999;
Dromi, Leonard, & Shteiman, 1993) but not for bilingual children in their L2 Hebrew
(Armon-Lotem 2012). Special attention was given to prepositions, which were found to better
discriminate SLI in Hebrew (Armon-Lotem, Danon, & Walters, 2008). The 56 items were
chosen from a larger pool of sentences which were tested on Russian-Hebrew bilingual
children with typical language development aged from 5;08 to 6;07 (Mean=6;01). All
children were born in Israel and grew up speaking Russian at home and Hebrew is the
preschool from age 3. All children scored within the monolingual norm in at least one of their
languages (Armon-Lotem, 2011). The pilot study showed that bilingual children with TLD
had difficulties with object clefts, passive constructions with the agent-phrase, and complex
sentences of dual complexity (e.g. questions containing a passive construction). Thus, it was
decided to eliminate these structures as they pose great difficulties to typically developing
bilingual children aged 6.
The 56 sentences selected vary in terms of embedding and syntactic movement,
including sentences without embedding and movement (simple mono-clausal sentences with
a range of inflections and prepositions), sentences with embedding but without movement
(bi-clausal sentences involving co-ordination vs. subordination), sentences with movement
but without embedding (object and oblique which-questions, VSO), and sentences with
embedding and movement (relative clauses). Following these principles made the task
comparable with the other languages. Bi-clausal sentences with advanced conjunctions as
well as conditionals were added to tap into semantic complexity. Eleven structures were
grouped into 3 blocks, as shown in Table 3, by length and complexity and pseudo-
randomized within each block so that two consecutive test items did not contain the same
target structure or the same lexical items. There were 8 sentences with SVO, the basic word
16
order for Hebrew, to enable independent testing of morphosyntax and prepositions and 10
more structures with 4-8 sentences in each. The same lexical items were deployed across
different syntactic structures to minimize the effect of specific lexical items. Stimuli
sentences varied in length from 5 to 11 words.
Table 3: Structures tested in the Hebrew SRep task with examples (number of sentences per
structure in brackets)
Block
Structure
Example
1
SVO (8)
ha- kelev maca et ha-ecem mitaxat la-mita
DET- dog found ACC DET-bone under to+DET-bed
‘The dog found the bone under the bed.’
Biclausal
Sentences with
coordination (4)
ha-yeladot oxlot glida ve ha-imahot šotot qafe.
DET-girls eat ice-cream and DET-mothers drink coffee.
‘The girls eat ice-cream and the mothers drink coffee.’
Biclausal
Sentences with
subordination (4)
ha-yeled xashav še- halaxta la- mesiba.
DET-boy thought that- went.2P.MASC.SG to+DET- party
‘The boy thought that you went to the party.’
2
Object Questions
(4)
eyze yeled ha-more hifxid
which.MASC boy DET-teacher.MASC frightened.3P.MASC.SG
etmol
yesterday
‘Which boy did the teacher frighten?’
Oblique
Questions (4)
le- eyzo mora ha-yalda hevi'a perax
to which.FEM teacher.FEM DET-girl brought.3P.FEM.SG flower
‘Which teacher did the girl bring a flower to?’
Object Relatives
(4)
ra'iti et ha-kelev še ha-sus daxaf
saw.1P.SG ACC DET-dog that DET- horse pushed.3P.MASC.SG
‘I saw the dog that the horse pushed.’
Oblique Relatives
(4)
ze ha-tinoq še ha-xatul meciq lo
this DET- baby that DET-cat annoy.3P.MASC.SG to+him
‘This is the baby that the cat annoys.’
3
Real Conditionals
(4)
im ha-boc yityabeš nece lesaxeq
if DET- mud dry.3P.MASC.SG.PRES go-out.2P.PL to+play
ba-xacer
in+DET-yard.
‘If the mud dries, we will go to play in the yard.’
Unreal
Conditionals (4)
im hayiti mištatef ba-meroc hayiti
if was.1P.SG take+part.PRES.MASC.SG in+DET-race was.1P.SG
zoxe
win.RES.MASC.SG
‘If I had taken part in the race, I would have won.’
Biclausal
lamrot še ha-yalda nafla hi
17
sentences with
advanced
Conjunctions (8)
even+thought that DET-girl fell.3P.FEM.SG she
himšixa ba-taxarut
continued.3P.FEM.SG in+DET competition
‘Even though the girl fell down, she continued (participating) in the
competition.’
VSO (8)
ba- boqer qar’a ha-yalda sefer
in+DET morning read.3P.FEM.SG DET-girl book.
‘In the morning the girl read a book.’
French SRep task
The French SRep task (Prévost, Tuller, & Zebib, 2012) has 56 items of varying complexity
and length. Special attention was given to language specific features of verbal morphology
which are central in French and known to be difficult for monolingual children with SLI,
focusing on present and past tense singular and plural. With French being a Romance
language, masculine and feminine object clitics were also targeted as these have been found
to be sensitive to SLI (Prévost, this volume). The 56 items were chosen from a larger pool of
sentences which were tested on bilingual children with TLD aged 5-6. The pilot study
showed poor performance of the bilingual children on a number of syntactic structures. Thus,
it was decided to eliminate these structures as they pose great difficulties to bilingual children
aged 6 with TLD. The 56 sentences selected vary in terms of embedding and syntactic
movement, including sentences without embedding and movement (simple mono-clausal
sentences), sentences with embedding but without movement (bi-clausal sentences involving
subordination with finite and non-fine sentential complements), sentences with movement but
without embedding (who and which object wh-questions, short and long passives, right
dislocation with clitics), and sentences with embedding and movement (relative clauses).
These structures made the task comparable with the other languages. Seven structures with 8
items each, as shown in Table 4, (some structures divided in two types with 4 sentences each,
such as subject and object relative clauses) were ordered by length and complexity and
pseudo-randomized so that two consecutive test items did not contain the same target
18
structure or the same lexical items. The same lexical items were deployed across different
syntactic structures to minimize the effect of specific lexical items. Stimuli sentences varied
in length from 4 to 10 words.
Table 4: Structures tested in the French SRep task with examples (number of sentences per
structure in brackets)
SVO, present tense (8)
Le bébé boit du lait.
the.MASC.SG baby drinks some.MASC milk
The baby drinks some milk.
SVO, passé compose (8)
La maman a fermé la fenêtre.
the.FEM mother has closed the window
The mother closed the window
Passives (8)
Short (4)
Ce matin, le cheval a été brossé.
this morning the.MASC.SG horse has been brushed
This morning, the horse was brushed
Long (4)
La dame a été poussée par le garçon.
the.FEM.SG woman has been pushed by the.MASC.SG boy
The woman was pushed by the boy
Accusative Clitics (Right
Dislocation) (8)
Je le remplis toujours ce verre.
I Cl.ACC.3sg.masc fill always this.SG.MASC glass
The glass, I always fill it
Biclausal
Sentences
with
subordination
(8)
Non-finite
sentential
complements
(4)
Le papa sait très bien conduire la voiture
the.MASC.SG daddy knows very well to+drive the.FEM car
The father knows to drive the car very well
Finite
sentential
complements
(4)
Le garçon dit que la maman a lu
the.MASC.SG boy says that the.FEM mommy has read
un livre.
a.MASC book
The boy says that the mother has read a book.
Relatives (8)
Subject (4)
Je vois la fille qui a poussé le garçon.
I see the.FEM girl that(subject) has pushed the.MASC boy
I see the girl that pushed the boy.
Object (4)
Je vois le garçon que la fille a poussé.
I see the.MASC.SG boy that(object) the.FEM girl has pushed
I see the boy that the girl pushed.
Object
Which-
questions (8)
Who (4)
Qui la dame dessine ?
who the.FEM lady draws
Who does the woman draw?
Which (4)
Quel enfant la maîtresse punit?
which child the.FEM teacher punishes
Which child does the teacher punish?
19
4. Procedure and scoring
The sentences in all languages were pre-recorded by native speakers of the language tested
and incorporated into PowerPoint presentations for administration of the tasks to ensure that
all children listen to the sentences in exactly the same way. PowerPoint presentations were
created for both the first and second stage of development of the SRep tasks. For the first
stage, a number represented each sentence and a smiley was used at the end of each group of
10 sentences to provide positive feedback and encourage participants to continue the task.
Figure 1 shows a slide from the Power Point of the English task that consisted of 60
sentences.
Figure 1: Slide from PowerPoint of the English SRep task (first stage, long SRep task)
For the languages, in which the task was organized by three blocks with raising
complexity, the three blocks were presented consecutively, as shown in Figure 1 (the colors
red, green and purple representing different blocks). For the languages that did not include
levels of complexity (e.g., French), all sentences were presented in one block.
20
For the second stage, the SRep tasks were embedded into a treasure hunt in order to
make the task more engaging. A bear was shown in a PowerPoint presentation going through
several locations in search of a treasure. Figure 2 provides an example with the bear going
towards a cave.
Figure 2: Slide from PowerPoint of the English SRep task (second stage, short SRep task)
In this version, the bear moved from stone to stone. On each stone the child heard a sentence
and had to repeat the sentence for the bear to move to the next one. This proved to be very
motivating for children who were eager to repeat the sentences in order to find out where the
bear would go and what the treasure was going to be.
The sentences were presented through headphones to ensure good and consistent
quality of input. External speakers were used for presentation in a very few cases in which
the child objected to the headphones. The children’s responses were recorded with high
quality microphones to allow detailed scoring after testing.
21
Children were seated in front of a computer screen showing the PowerPoint
presentation. They were instructed to listen carefully to the sentences and say exactly what
they heard. A practice session presented 2 to 4 sentences to demonstrate the task to the
children and to ensure that the children understood what they had to do. If children did not
repeat the sentences spontaneously, the experimenter explained the task again and
encouraged them to repeat what they heard. The practice sentences could be repeated several
times until the children understood what they had to do. Most children understood the task
quickly and easily.
For the experimental sentences, children listened to each sentence only once unless
there was a loud noise in the room or another interruption. Oral praise (´well done´) was used
irrespective of whether or not the children were repeating the sentences accurately in order to
motivate the children to continue. If the child self-corrected, their final response was scored
(whether correct or incorrect).
The children`s responses were audio recorded and marked on an answer sheet during
the testing. They were further transcribed and scored off-line on the basis of the recording.
Figure 3 illustrates the first page of the scoring sheet of the English SRep task.
22
Figure 3: Scoring sheet of English 60 item SRep task
Sentence repetition tasks can be scored in different ways depending on whether they are
being used for research or clinical purposes, the focus of the study, and practical reasons,
23
such as the amount of time available. For the LITMUS-SRep tasks we initially used six
scoring schemes and compared them with each other at the pilot phase of the studies.
The first two schemes, 0-1 and 0-3 schemes, calculate sentence scores (column S in
the scoring sheet in Figure 3). The 0-1 scheme allocated a score of 1 if the sentence was
repeated entirely verbatim and a score of 0 if there were one or more changes in the
children’s response. This scoring was based on the Test of Language Development-Primary
(TOLD-P-4) (Newcomer & Hammill, 2008). The 0-3 scheme allocated a score of 3 if the
sentence was repeated entirely verbatim, a score of 2 if there was one change, a score of 1 if
there were 2 to 3 changes and a score of 0 if there were 4 or more changes in the children’s
response. This scoring was based on the CELF-3 (Semel et al., 1995).
The third scheme counted the number of content words (Content word score) and
function words (Function word score) correct and noted the number of omissions,
substitutions, and additions. It was used to explore differences between word and error types.
To facilitate scoring, content words were highlighted blue and function words were
highlighted yellow, as shown in Figure 3. This scoring scheme was based on the Sentence
Imitation Test (SIT) (Seeff-Gabriel, Chiat, & Roy, 2008).
The fourth was a grammaticality scheme. It allocated 1 point if the sentence had no
grammatical errors, regardless of whether or not it was a verbatim repetition of the target
sentence and a mark of 0 if the sentence had one or more grammatical errors.
The fifth scheme was a sentence structure score. It allocated 1 point if the child used
the structure targeted irrespective of whether or not there were changes in other parts of the
sentence. A score of 0 was given if the child made an error in the sentence structure that was
targeted, e.g. omission of an auxiliary in sentences targeting auxiliaries. A score of 0 was also
given if the child did not produce the sentence structure targeted and substituted it with
another structure, e.g. substituting an object relative clause (The swan that the deer chased
24
knocked over the plant) with a subject relative clause (The swan that chased the deer knocked
over the plant) or substituting a passive (The child was helped in the sweet shop) with an
active (The child was helping in the sweet shop). The score sheet indicated the structure that
was targeted in the column ‘syntactic structure’.
The sixth scoring scheme calculated the number of changes between the target and the
repeated structure using the Levenshtein Distance algorithm (Levenshtein, 1966) which
measures the distance between the target and the response. This scheme was based on Riches
et al. (2010) and was calculated automatically in an excel spreadsheet.
In all schemes, allowances were given for self-corrections, consistent phonological
processes, e.g. fronting, dialectal variation, contractions (they’re for they are), and
expansions (John will not for John won’t).
5. Pilot data
ENGLISH. A pilot study of the English long SRep task was conducted with 100 6-to8-year
old children (60 monolingual children with TLD, 20 Polish-English children with TLD, and
20 monolingual English children with SLI). All scoring schemes were used in order to
identify which scoring schemes differentiated maximally the groups of children with TLD
from the group of children with SLI. The 0-1, 0-3, and the grammaticality scheme were best
in separating the monolingual and bilingual children with TLD from the children with SLI. In
terms of the structures targeted, the monolingual and bilingual children with TLD
outperformed the children with SLI in all sentence types, but some structures showed no
overlap between the children with SLI vs. the groups with TLD, whereas some other
structures showed partial overlap because they were difficult also for the monolingual and
bilingual children with TLD. The structures showing no overlap between the monolingual
25
and bilingual children with TLD on the one hand and the children with SLI on the other hand
were the SVO sentences with one auxiliary/modal, two auxiliaries or one auxiliary and one
modal, the short and long passives, the who/what wh-questions, the right branching object
relative clauses, the sentences with conditionals, and the sentences with nouns taking
complements. SVO sentences with negation and an auxiliary/modal, object clefts and subject
clefts with passives were difficult for all groups of children and as a result showed overlap
between the two groups of children with TLD and the children with SLI. Finally, some
structures were difficult for the bilingual children with TLD and as a result showed overlap
between this group and the group of children with SLI. These were the bi-clausal sentences
with coordination and infinitival clauses, the sentences with negation one auxiliary and one
modal, which-object and oblique object questions, bi-clausal sentences with subordination,
and centre embedded object relative clauses.
RUSSIAN-HEBREW. In a pilot study, Meir (2010) tested 12 bilingual children with TLD (6
female) aged from 5;6 to 6;5 (Mean = 6;08) in L1 Russian and L2 Hebrew. All children were
born in Israel. Two scoring methods were applied: correct full sentence (1- correct repetition,
0- erroneous repetition) and correct target structure (1- correctly preserved target structure, 0
– target structure was modified / omitted). The pilot identified the syntactic structures in
which bilingual children with TLD performed poorly (e.g. passives with the agent-phrase in
Russian and Hebrew, object clefts in Hebrew) and helped to decide which structures should
be eliminated from the current version of the Russian and Hebrew SRep tasks on the grounds
that they posed difficulties for bilingual children with TLD. The Russian and Hebrew tasks
were administered to 38 sequential bilingual children with L1 Russian and L2 Hebrew (10
with SLI) (Meir et al., under review). The bilingual group with TLD significantly
outperformed the bilingual SLI group on most morpho-syntactic structures in both languages.
26
In L1 Russian, children with TLD showed better performance than the children with SLI on
sentences with the VSO, OVS, SOV word orders, biclausal sentences with coordination and
subordination, oblique questions, subject and object relatives and real conditionals. However,
object questions and unreal conditionals posed difficulties for all children. In Hebrew, the
group with TLD repeated sentences containing object and oblique questions, object and
oblique relatives, sentences with VSO word order, real and unreal conditionals and biclausal
sentences with advanced conjunctions more accurately than their SLI peers. However, both
groups showed a ceiling effect on sentences with SVO word order and biclausal sentences
with coordination and subordination. Moreover, four error patterns distinguished the SLI
group from the group with TLD: avoiding a response all together, omitting coordinators,
subordinators and prepositions and using simple clauses rather than relative clauses.
6. Discussion of strengths, limitations and recommendations for using the
LITMUS-SRep tasks
SRep tasks have several advantages compared to other types of tasks. They are quick and
easy to administer, they have clear target sentences and can include a large range of sentence
types, and they can be scored in several ways depending on the focus of the analysis (Gabriel,
Chiat, & Dodd, 2010).
The LITMUS-SRep tasks have all these strengths. In terms of the length of the tasks,
the first version of the LITMUS-SRep tasks included approximately 60 sentences for each
language including a large range of sentence structures. The task took around 20 to 30
minutes to administer. There is a tension between the length of a task and how
comprehensive it is in providing data for a range of structures. Long SRep tasks can be less
engaging for children and less appropriate within a clinical setting. Therefore, shorter
27
versions of the LITMUS-SRep tasks including approximately 30 sentences were created
based on pilot data. The shorter tasks take approximately 10 minutes to administer. The tasks
were shortened on the basis of inter-item reliability and internal consistency measures, but
also empirical evidence about which structures are better in differentiating between
monolingual and bilingual children with TLD and children with SLI.
The procedure of the LITMUS-SRep tasks was standardized between languages using
pre-recorded sentences and the same PowerPoint slides in both stages across languages.
Therefore, differences between languages cannot be attributed to differences in the task
procedure. Scoring was also standardized across languages and includes a range of scoring
schemes. These scoring schemes provide quantitative scores that can be used to compare the
participants’ performance between structures and between languages. In addition, qualitative
analyses can be conducted by analyzing the error patterns. Qualitative analyses can be
tailored to research and/or clinical needs. They can provide invaluable information about the
cause of the breakdown in repeating sentences and can be extremely helpful when designing
language intervention protocols.
SRep tasks also have important limitations that need to be kept in mind. The most
important limitation concerns the relationship between the developmental stage of the
participant and the sentence length. If sentences are too short for the participant’s language
abilities, they may be able to repeat them passively (parroting) and the results will not reflect
the participant’s language abilities. To avoid this limitation, the LITMUS-SRep sentences are
relatively long. However, the opposite effect may also occur. If the sentences are too long,
they may exceed the participant’s memory capacity and may lead to a floor effect.
Participants may fail to repeat sentences not because they have not acquired a particular
structure, such as auxiliaries in the English task, but for different reasons, for example
because of the length of the sentence or the lexical items used. This can be partially addressed
28
through the structural scoring, which identifies if participants were able to use the specific
structure irrespective of whether or not they repeated the sentence verbatim.
A second important limitation is that it does not assess all aspects of language ability
and all levels of representation. While it is extremely informative about lexical,
morphosyntactic, and syntactic skills, it does not separate comprehension from production
and it does not assess the children’s pragmatic skills. Therefore, despite its strengths, it
should be supplemented by other language assessments.
When the LITMUS-SRep task is administered to bilingual children, it should be used
in conjunction with a detailed language history questionnaire, such as the PABIQ
questionnaire (see Tuller, this volume). This is crucial because differences in the participants’
language history, such as AoO, LoE, quantity and quality of input may affect the participants’
performance on the LITMUS-SRep tasks in the two languages. For example, a very short
LoE in one of the two languages may result in a very low performance in that specific
language. In fact, it would not be recommended to use the SRep task in a specific language if
the LoE in a specific language is less than 12 months since we would expect performance to
be very low with very limited LoE. Furthermore, the task may be frustrating for children with
such limited exposure.
The tasks should be administered following the protocols developed. When both
languages spoken by the child are tested, these should be tested in separate sessions and
instructions should be given in the language tested. In each SRep task, instructions are
included in the PowerPoint presentation. The task protocols should be followed carefully on
the basis of the task manuals. Giving allowances consistently for self-corrections,
phonological processes, and expansions is crucial, as well as calculating interrater reliability
scores for transcription and scoring. Finally, it is very important to keep the children engaged
29
in the task by praising them for their responses irrespective of whether or not they repeat the
sentences verbatim.
Acknowledgements
The sentence repetition tasks were developed within the COST Action IS0804 ‘Language
Impairment in a Multilingual Society’. At the time of writing, Theodoros Marinis was
supported by an Onassis Fellowship and Sharon Armon-Lotem was supported by the
German-Israel Foundation (Grant No. 1113/2010). We would like to thank all members of
the Sentence Repetition group for developing tasks in a range of languages and Vicky
Chondrogianni, Enkeleida Kapia, and Aneta Miekisz for working on the PowerPoint
presentation of the short SRep task.
30
References
Adani, F., van der Lely, H. K., Forgiarini, M., & Guasti, M. T. (2010). Grammatical feature
dissimilarities make relative clauses easier: A comprehension study with Italian
children. Lingua, 120(9-3), 2148-2166. doi: 10.1016/j.lingua.2010.03.018
Armon-Lotem, S. (2011). Design and Data of English-Hebrew and Russian-Hebrew
Sentence Repetition Tasks. Paper presented at the Paper presented at COST IS0804
WG and MC Meeting, Malta, November.
Armon-Lotem, S. (2012). Between L2 and SLI: Inflections and prepositions in the Hebrew of
bilingual children with TLD and monolingual children with SLI. Journal of Child
Language(41), 1-31.
Armon-Lotem, S., Danon, G., & Walters, J. (2008). The use of prepositions by bilingual SLI
children: The relative contribution of representation and processing Proceedings of
the Generative Assembly on Language Acquisition (pp. 41-46 ).
Armon-Lotem, S., Gagarina, N., & Walters, J. (2011). The impact of internal and external
factors on linguistic performance in the home language and in L2 among Russian-
Hebrew and Russian-German preschool children. Linguistic Approaches to
Bilingualism, 1, 291-317.
Baayen, R. H., Piepenbrock, R., & Gulikers, L. (1995). The CELEX Lexical Database (CD-
ROM). Paper presented at the Linguistic Data Consortium, University of
Pennsylvania, Philadelphia, USA.
Baddeley, A. (2000). The episodic buffer: a new component of working memory? Trends
Cogn Sci, 4(11), 417-423. doi: S1364-6613(00)01538-2 [pii]
Baddeley, A. (2012). Working memory: theories, models, and controversies. Article in
Annual Review of Psychology, 63, 1-19.
Bley-Vroman, R., & Chaudron, C. (1994). Elicited imitation as a measure of second-language
competence. In T. E. E, G. S. M & C. A. D (Eds.), Research methodology in second-
language acquisition (pp. 245-262). Hillsdale, NJ: Lawrence Erlbaum.
Boyle, W., Lindell, A. K., & Kidd, E. (2013). Investigating the role of verbal working
memory in young children’ssentencecomprehension. LanguageLearning, 63, 211–
242.
Chiat, S., Armon-Lotem, S., Marinis, T., Polisenska, K., Roy, P., & Seeff-Gabriel, B. (2013).
Assessment of language abilities in sequential bilingual children: the potential of
sentence imitation tasks. In V. C. Mueller Gathercole (Ed.), Bilinguals and
assessment: State of the art guide to issues and solutions from around the world (pp.
56-89). Bristol: Multilingual Matters.
Conti-Ramsden, G., Botting, N., & Faragher, B. (2001). Psycholinguistic markers for specific
language impairment. Journal of Child Psychology and Psychiatry, 6, 741-748.
Daniel, D. B., & Klaczynski, P. A. (2006). Developmental and individual differences in
conditional reasoning: effects of logic instructions and alternative antecedents. Child
Development, 77, 339-354.
Dromi, E., Leonard, L., Adam, G., & Zadunaisky-Ehrlich, S. (1999). Verb agreement
morphology in Hebrew-speaking children with Specific Language Impairment.
Journal of Speech, Language and Hearing Research, 42, 1414-1431.
Dromi, E., Leonard, L., & Shteiman, M. (1993). The grammatical morphology of Hebrew
speaking children with specific language impairment: some competing hypotheses.
Journal of Speech and Hearing Research, 36, 760-771.
Ellis, R. (2005). Measuring implicit and explicit knowledge of a second language: A
psychometric study. . Studies in Second Language Acquisition, 27, 141–172.
31
Erlam, R. (2006). Elicited imitation as a measure of L2 implicit knowledge: An empirical
validation study. Applied Linguistics, 27, 464–491.
Fattal, I., Friedmann, N., & Fattal-Valevski, A. (2011). The crucial role of thiamine in the
development of syntax and lexical retrieval: a study of infantile thiamine deficiency.
Brain, 134, 1720-1739.
Fearnley, S. (1997). MRC Psycholinguistic Database search program. Behavior Research
Methods, Instruments, & Computers, 29, 291-295
Friedmann, N., & Novogrodsky, R. (2004). The acquisition of relative clause comprehension
in Hebrew: A study of G-SLI and normal development. Journal of Child Language,
31, 661-681.
Friedmann, N., & Novogrodsky, R. (2011). Which questions are most difficult to understand?
The comprehension of Wh questions in three subtypes of SLI. Lingua 2011, 121,
367–382.
Gabriel, B., Chiat, S., & Dodd, B. (2010). Sentence imitation as a tool in identifying
expressive morphosyntactic difficulties in children with severe speech difficulties.
International Journal of Language and Communication Disorders, 45, 691-702.
Gutierrez-Clellen, V. F., Restrepo, M. A., & Simon-Cereijido, G. (2006). Evaluating the
discriminate accuracy of a grammatical measure with Spanish-speaking children.
Journal of Speech, Language and Hearing Research, 49, 1209-1223.
Jessop, L., Suzuki, W., & Tomita, Y. (2007). Elicited Imitation in Second Language
Acquisition Research. The Canadian Modern Language Review, 64, 215–238.
Leonard, L. (1998). Children with specific language impairment. Cambridge, Mass: MIT
Press.
Levenshtein, V. I. (1966). Binary codes capable of correcting delections, insertions, and
reversals. Soviet Physics — Doklady, 10, 707–710.
Lombardi, L., & Potter, M. C. (1992). The regeneration of syntax in short term memory.
Journal of Memory and Language, 31, 713–733.
Lust, B., Flynn, S., & Foley, C. (1996). What children know about what they say: Elicited
imitation as a research method for assessing children's syntax. In D. McDaniel, C.
McKee & S. Cairns Helen (Eds.), Methods for assessing children's syntax (pp. 55-76).
Cambridge, MA: MIT Press.
Marinis, T., Chiat, S., Armon-Lotem, S., Gibbons, D., & Gipps, E. (2010). School-Age
Sentence Imitation Test (SASIT) University of Reading. Reading.
Marinis, T., & Saddy, D. (2013). Parsing the Passive: Comparing Children With Specific
Language Impairment to Sequential Bilingual Children. Language Acquisition, 20(2),
155-179. doi: Doi 10.1080/10489223.2013.766743
Marshall, C., Marinis, T., & van der Lely, H. (2007). Passive verb morphology: The effect of
phonotactics on passive comprehension in typically developing and Grammatical-SLI
children. Lingua, 117, 1434-1447.
Meir, N., & Armon-Lotem, S. (2014). Disentangling bilingualism from SLI in Heritage
Russian: The impact of L2 properties and length of exposure to the L2 GALA
Proceedings.
Meir, N., Walters, J., & Armon-Lotem, S. (under review). Bidirectional morpho-syntactic
transfer: the role of language proficiency and length of exposure. International
Journal of Blingualism.
Montgomery, J., & Evans, J. (2009). Complex sentence comprehension and working memory
in children with Specific Language Impairment. Journal of Speech, Language, and
Hearing Research, 52, 269-288.
32
Newcomer, P., & Hammill, D. D. (2008). Test of Language Development: Primary, 4th
edition. Austin, TX: Pro-ed.
Paradis, J. (2005). Grammatical morphology in children learning English as a Second
Language: Implications of similarities with Specific Language Impairment.
Language, Speech and Hearing Services in the Schools, 36, 172-187.
Polinsky, M. (2006). Incomplete acquisition: American Russian. Journal of Slavic
Linguistics, 14, 191-262.
Polišenská, K. (2011). The influence of linguistic structure on memory span: repetition tasks
as a measure of language ability. City University London.
Polišenská, K., Chiat, S., & Roy, P. (under review). The contribution of linguistic knowledge
to immediate verbal recall in children.
Potter, M. C., & Lombardi, L. (1990). Regeneration in the short-term recall of sentences.
Journal of Memory and Language, 29, 633–654.
Potter, M. C., & Lombardi, L. (1998). Syntactic priming in immediate recall of sentences.
Journal of Memory and Language, 38, 265-282.
Prévost, P., Tuller, L., & Zebib, R. (2012). LITMUS-SR-French. Tours.
Ravid, D., Landau, H., & Lovetski, M. (2003). Passive morphology in the school years: An
example of later language acquisition (in Hebrew). Script, 5-6, 129-158.
Rice, M. L., & Wexler, K. (1996). Toward tense as a clinical marker of specific language
impairment in English-speaking children. Journal of Speech and Hearing Research,
39, 1239-1257.
Riches, N. G., Loucas, T., Baird, G., Charman, T., & Simonoff, E. (2010). Sentence
repetition in adolescents with specific language impairments and autism: an
investigation of complex syntax. International Journal of Language and
Communication Disorders, 45, 47-60.
Seeff-Gabriel, B., Chiat, S., & Roy, P. (2008). Early Repetition Battery. London: Pearson
Education, Inc.
Semel, E., Wiig, E., & Secord, W. A. (1995). Clinical Evaluation of Language Fundamentals
3 (CELF-3). San Antonio: The Psychological Corporation.
Slobin, D. I., & Welsh, C. A. (1973). Elicited imitation as a research tool in developmental
psycholinguistics. In C. A. Ferguson & D. I. Slobin (Eds.), Studies in child language
development. New York, NY: Holt, Rinehart, & Winston.
Thordardottir, E., & Brandeker, M. (2013). The effect of bilingual exposure versus language
impairment on nonword repetition and sentence imitation scores. Journal of
Communication Disorders, 46, 1-16.
van der Lely, H. K. J. (1996). Specifically language impaired and normally developing
children: Verbal passive vs. adjectival passive sentence interpretation. Lingua, 98,
243-272.
Verhoeven, L., Steenge, J., & van Vanbalkom, H. (2012). Linguistic transfer in bilingual
children with specific language impairment. International Journal of Language and
Communication Disorders, 47, 176–183.