DEVELOPMENTAL DISABILITIES

RESEARCH REVIEWS 17:160169 (2011)

EARLY LANGUAGE DELAY AND SPECIFIC

LANGUAGE IMPAIRMENT
Jayne Moyle, Stephanie F. Stokes,* and Thomas Klee
Department of Communication Disorders, University of Canterbury, New Zealand

Early language delay (ELD) is a warning sign that may presage the
presence of a later language impairment (LI). In order to allow more tar-
geted identification and earlier intervention for LI, better diagnostic
measures for toddlers are needed. Development of accurate predictive/
diagnostic models requires consideration of a set of complex interrelated
questions around definition, causality, and theories of LIs. A multifacto-
rial model of language development and LI is essential to increase the
accuracy of prediction. This article examines what is known about LI in
the preschool years and language delay in toddlers, and examines these
in relation to the Procedural Deficit Hypothesis (Ullman and Pierpont,
[2005] Cortex 41:399433] and the Statistical Learning Account (Stokes
et al., [2012a] J Speech Lang Hear Res; Stokes et al., [2012b] J Child
Lang 39:105129) to suggest a new framework for characterizing ELD
to better assist prediction of later LI. VC 2013 Wiley Periodicals, Inc.
one of the challenges in early identification of this disorder.
The focus of this article is the developmental period in which
a core language disorder is identified. Language disorders in
children are usually identified at four-to-five years of age.
However, there are indications that SLI might be detectable
earlier than the age at which it is usually diagnosed.
LATE TALKING AND LATER LANGUAGE
IMPAIRMENT
Late onset of talking has long been thought to be an
early manifestation of later LI [Olswang et al., 1998; Thal,

Dev Disabil Res Rev 2012;17:160169.

Key words: late talkers; preschool specific language impairment
2000] and has been referred to variously as specific expressive
language delay (SELD), early language delay (ELD), and late
language emergence (LLE). It is identified between about 18
and 30 months of age, with the first indicators being early vo-
cabulary delay, slow vocabulary growth, and/or delayed onset
of word combinations. For research purposes, this group of

C
hildren with language disorders experience difficulties
children has been identified on the basis of parent-report
with communication which may impact adversely on
measures such as the Language Development Survey [LDS;
social-emotional development, academic achievement,
Rescorla, 1989], the MacArthur Communicative Develop-
and vocational options if the disorder persists beyond the pre-
ment Inventory: Words and Sentences [CDI:WS; Fenson
school period [Clegg et al., 2005]. During this period, a
et al., 2007], and the Ages and Stages Questionnaire Commu-
language impairment (LI) in the absence of significant sensory,
nication scale [ASQ; Bricker and Squires, 1999]. Various
psychiatric, neurological, or intellectual impairment/disorder is
operational definitions have been used to define the target
commonly referred to as specific language impairment (SLI).
population, including (1) fewer than 50 words of expressive
While the profile of specific linguistic abilities (e.g., in gram-
vocabulary or no word combinations by 24 (6 6) months of
mar, semantics, phonology, and pragmatics) varies across
age on the LDS [e.g., Rescorla, 1989; Klee et al., 1998]; (2)
children with SLI and across time, English-speaking children
<10th percentile on the expressive vocabulary section of the
with SLI generally exhibit problems with grammatical mor-
CDI:WS [e.g., Reilly et al., 2007]; and (3) 1 SD or more
phology relative to children of the same age without SLI
below the mean on a composite score from the ASQ Com-
[Leonard, 1998]. In an effort to understand this variability across
munication scale [Zubrick et al., 2007]. These definitions
linguistic abilities, researchers have identified subgroups of LI.
typically result in between 10 and 20% of children being
One attempt to delineate subgroups of children with
identified as late talkers (LTs).
SLI using cluster analysis found three underlying linguistic fac-
While these criteria are commonly used, they appear to be
tors based on phonology/articulation, semantics/syntax, and
at best only moderately indicative of ongoing language disorder.
pragmatics [Tomblin et al., 2004]. These factors dissociate to
This is likely due to the wide variation in the development of
some degree to create different subgroups. Further, semantic/
early language skills such as vocabulary and grammar. Late talk-
syntactic disorders may be expressive only or both receptive
ing has been associated with poorer developmental outcomes
and expressive. While there is as yet no consensus regarding
how many subgroups exist, it is clear that a childs subgroup
membership and developmental profile can change over time
[Conti-Ramsden and Botting, 1999]. There is evidence to *Correspondence to: Stephanie F. Stokes, Department of Communication Disor-
ders, University of Canterbury, Private Bag 4800, Christchurch, 8140, New Zea-
suggest that these variations in how the disorder manifests land. E-mail: stephanie.stokes@canterbury.ac.nz
itself may be characteristic of the core language disorder as it Received 11 September 2012; accepted 5 October 2012
unfolds over time, rather than separate diagnostic entities View this article online in Wiley Online Library (wileyonlinelibrary.com).
DOI: 10.1002/ddrr.1110
[Bishop, 1994]. The heterogeneity of the SLI population is
' 2013 Wiley Periodicals, Inc.
such as SLI, dyslexia, autism spectrum
disorders, and attention deficit hyperac-
tivity disorder [Buschmann et al., 2008].
In general, children who start out as high
achievers in language will continue to
achieve well throughout childhood
whereas those who start off slowly often,
but not always, have poorer outcomes.
However, the majority of children who
start off slowly will accelerate in their de-
velopment to come within normal range
over the next 18 months [Paul, 1996;
Rescorla and Roberts, 1997]. In compar-
ison with the 5070% recovery rate from
late talking, a diagnosis of SLI at age 4 has
a smaller degree of recovery in the short-
term. Bishop and Edmundson [1987]
found that of children with SLI at age 4
identified by a narrative measure, only
10% had resolved by age 56. In a longi-
tudinal study with the same sample,
Stothard et al., [1998] found that children
who had resolved by age 56 differed
from controls at ages 1516 years on
measures of phonological processing and
literacy. However, those who had SLI or
a general delay at age 56 had significant
impairments in all aspects of spoken and
written language functioning at out-
come. This illustrates the changing
nature of language delay and disorder
across childhood. As a group, children
with an early vocabulary delay fall within
the range of typical development by 45
years but then delays in grammatical de-
velopment become apparent.
Conversely, there is some evidence that a
portion of children start out in the nor-
mal range for language but slow in
development, and fall in the disordered
range at a later point [Dale et al., 2003;
Rice et al., 2008]. Predicting which LTs
will go on to manifest a significant lan-
guage difficulty at age 45 poses a
diagnostic challenge (see Rescorla, this
issue), with part of the problem being
what exactly constitutes SLI.
While the term specific is part of
the definition, non-linguistic skills are
also known to be impaired, with chil-
dren having depressed nonverbal IQ
scores [Johnston, 1994], and deficits in
hypothesis testing [Kamhi et al., 1984],
cross-modal processing [Montgomery,
1993], voice processing abilities [Creu-
sere et al., 2004], sustained attention
[Finneran et al., 2009], processing
capacity and speed [Leonard et al.,
2007], working memory [Montgomery
et al., 2010], and motor coordination
[Hill, 2001]. This begs the question of
whether or not children with more
general developmental delays are
included in studies predicting SLI.
Commonly, children with general de-
DEV DISABIL RES REV  EARLY LANGUAGE DELAY
velopmental delays [indicated by
nonverbal IQ scores below 85] have
not been included. However, this prac-
tice is questionable for several reasons.
Firstly, the assumption behind this
appears to be that children with low
nonverbal scores do not have the cogni-
tive underpinnings to support normal
language development; that is, that an
undetermined general factor which
caused low nonverbal achievement will
have also caused the problem with lan-
guage development. However, groups of
children with low nonverbal ability but
language in the normal range have been
identified [Rice et al., 2004]. The lan-
guage ability of these children did not
differ from age-matched controls. Sec-
ondly, it is debatable whether or not the
language of children with SLI is qualita-
tively different from those with more
general delays. Rice et al. [2004]
reported qualitative differences while
Tomblin et al. [2004] reported only
quantitative differences. Children on the
borderline between these two diagnoses
shift between them over time, pointing
to a degree of overlap in these clinical
groups [Vig et al., 1987], leading some
researchers to prefer the term primary
language impairment (PLI) or language
impairment (LI) rather than SLI. In this
review, we adopt the term SLI as there
is still significant debate about comorbid-
ity and specificity of language-related
disorders. Thirdly, studies of other de-
velopmental disorders are likely to be
informative with respect to the defini-
tion of SLI. Boundaries between
disorder types may not be distinct. For
example, an overlap between dyslexia
and SLI has been postulated, with the
shared variance thought to be due to an
underlying phonological processing dis-
order [Catts et al., 2004]. In addition,
the categories of autism spectrum disor-
der and SLI may overlap to some extent
because of an underlying auditory per-
ceptual processing difficulty [Oram
Cardy et al., 2008]. There is a moderate
rate of comorbidity (or arguably overlap)
with SLI and childhood apraxia of
speech, attention deficit hyperactivity
disorder (ADHD), and autism. Diagnos-
ticians attempt to identify clusters of
children who fit these categories; how-
ever there are always individuals who do
not fit clear diagnostic categories, and
who sit at the boundary between types.
Children initially identified as LTs may
later be diagnosed with SLI, dyslexia,
ADHD, autism spectrum disorder
(ASD), or general intellectual disability,
moving in and out of diagnostic catego-
ries as they mature.
AND SLI  MOYLE ET AL.
PREDICTING LATER SLI
We turn our attention now to fac-
tors that may contribute to an ELD (and
thus, later talker status), as well as a pre-
school LI. A range of linguistic,
demographic, medical, genetic, and envi-
ronmental factors have been explored as
predictors of later language ability
[Zubrick et al., 2007; Ellis and Thal,
2008; Reilly et al., 2009; Stokes and Klee,
2009a]. While late emergence of expres-
sive vocabulary alone is not a strong
predictor of language ability at 4 or 5
years, other variables have some, but lim-
ited, predictive value. Language
comprehension and gestural use have
been shown to have some predictive value
in small sample studies [e.g., Thal et al.,
1991; Thal and Tobias, 1992]. Of the 10
LTs included in these studies, six were
late bloomers (i.e., children who had
caught up with their age-matched peers)
by 3035 months, while four were still
delayed in language production. Exami-
nation of their gesture use and language
comprehension at 18 months showed that
LTs who continued to have poorer lan-
guage skills had lower comprehension
scores, and failed to use gestures as com-
pensation for reduced expressive
vocabulary ability, than the late bloomers.
Ellis and Thal [2008], in a brief review of
the relationship between ELD and later
LI, suggested that there were three key
likely predictors, a family history of lan-
guage delay or impairment, delays in
receptive as well as expressive vocabulary,
and the lack of communicative gestures
(the latter is supported by a longitudinal
study [Rowe et al., 2012]). To these they
add a history of otitis media, parent con-
cern, high risk environments (e.g.,
poverty), and delays in symbolic play or
social skills. Mothers education [Dolla-
ghan et al., 1999], a family history of
speech/language delay [Bishop et al.,
2003], the childs sex, and early neurobio-
logical growth [Zubrick et al., 2007],
birth order/number of siblings [Stokes,
1997], a history of ear infections or parent
concern [Klee et al., 2000], socio-emo-
tional development [Bornstein et al.,
1998], and cognitive development [Price
et al., 2003], have been shown to have
some predictive value, in some, but not
all, studies. Evidence for these factors as
sole predictors is not consistent, and, as
we show below, a multifactorial model of
prediction is likely to be the way forward
for improving our ability to predict the
consequences of ELD.
A MULTIFACTORIAL MODEL
As we work toward a greater
understanding of factors that contribute
161
 Figure 1 Processes, components, and modalities of language development (from Klee and Stokes, [2011]).
to late talking and SLI, we must con-
sider multiple interrelated factors
[Desmarais et al., 2008]. Klee and
Stokes [2011] presented a multifactorial
model of language development that
draws on Gottliebs model of epigenesis
[Gottlieb, 2007], and Bishop and
Snowlings [2004] model of relation-
ships between observed behavior,
cognitive processes, neurobiology, and
etiology in development (see Fig. 1).
The model considers the roles of
genetic, environmental, neurobiological,
and cognitive factors in language devel-
opment. Here, we consider how these
factors might serve as predictors of later
SLI. For example, while the genotype
is the starting point for development,
identical genotypes can result in differ-
ent phenotypic outcomes in response to
differences in the environment (for
example, the variations seen in mono-
zygotic twin development).
FACTORS POTENTIALLY
PREDICTIVE OF SLI
Genetic Factors
Genetic factors play an important
role in the development of SLI. Verbal
ability seems to be one of the most her-
itable of cognitive traits, as demonstrated
by adoption studies [Plomin et al.,
1997]. Twin studies have shown that
monozygotic twins have higher rates of
concordance of language disorders than
dizygotic twins, indicating the impact of
genes beyond that of the environment
[Dale et al., 2003]. LTs are more likely
162
to have a family history of language and
learning disorders than children with
non-affected language development
[Zubrick et al., 2007; Reilly et al.,
2009]. They are also more likely to
score lower as a group on linguistic
related measures [Flax et al., 2009] and
to have SLI [Choudhury and Benasich,
2003]. In addition, boys are more likely
to be LTs than girls [Zubrick et al.,
2007; Reilly et al., 2009]. Boys show
slower expressive vocabulary growth
patterns than girls before 24 months, but
there is no gender difference in overall
vocabulary growth [Rowe et al., 2012].
The search for specific genes
causing neurodevelopmental disorders
has proven more complex than initially
expected. The concept of a language
gene is an appealing but simplistic
one. Early studies of the genotype of
the KE family, who have a high rate of
childhood apraxia of speech in their
family, yielded evidence that the
FOXP2 gene was deficient for affected
family members [Vargha-Khadem et al.,
1998]. Recent research suggests that
mutations of FOXP2, a transcription
factor gene that encodes a regulatory
protein, are rare and that it is not such
mutations that are relevant for SLI,
rather, it the endpoint of FOXP2 regu-
lation, the target neural circuitry that is
relevant [Fisher and Scharff, 2009].
FOXP2 regulates regions of the brain
thought to play major roles in the
cognitive mechanisms employed in lan-
guage learning, including the cerebral
cortex, basal ganglia, and thalamus.
DEV DISABIL RES REV  EARLY LANGUAGE DELAY
Vernes et al. [2008] reported that var-
iants in CNTNAP2, a gene regulated
by FOXP2, were related to poor per-
formance on language measures in
school-aged children with SLI. Further,
and relevant here, Whitehouse et al.
[2011], in a study of 1,149 2-year-old
children, suggested that variations in
CNTNAP2 may influence early lan-
guage development. Whitehouse et al.
[2011] proposed that this genetic var-
iance could increase the risk of
development of SLI, but added the ca-
veat that this was unlikely to occur in
isolation of other genetic or environ-
mental influences [see Bishop (2009),
for a review]. At this stage, for clinical
purposes, all we can say is that a family
history of language or learning disorders
should be considered a risk factor when
predicting forward from an ELD, but a
family history alone is not sufficient as a
predictive factor.
Environmental Factors
While genetic factors are critical
in the phenotypic outcome of disorders,
environmental stimulation also plays a
powerful role. Children learn language
from social interaction with other peo-
ple. Research into the effects of the
environment on language acquisition
has focused on two factors, conversa-
tional styles of parentchild interactions
and variations in linguistic input. The
role each plays is likely to differ as a
function of age and also at the extreme
ends of the ability and environment
spectra.
AND SLI  MOYLE ET AL.
 Some aspects of parentchild
interaction have been found to facilitate
language development [Yoder and
Warren, 2004]. Several studies have
examined whether the interactions
between LTs and their parents are dif-
ferent in a way that may negatively
impact on child language development.
However, the interaction styles of
parents of typically developing [TD]
children and parents of LTs appear to
be more similar than different. Paul and
Elwood [1991] reported that parents of
LTs have a larger gap between their
mean length of utterance and their
childs language, which may make lan-
guage more difficult to learn. Vigil
et al. [2005] found that parents of TD
children used more responses, expan-
sions, and self-directed speech than
parents of LTs; however these behaviors
could be reflective of higher language
abilities in TD children, which elicit
different conversational styles, rather
than style differences causing LLE.
Although the amount of linguistic input
experienced by children from different
social groups has been shown to have a
dramatic impact on the size of child-
rens developing lexicons [Hart and
Risley, 1995], the more subtle differen-
ces in the features of parentchild
interactions of late talking and TD chil-
dren are not likely to play a causal role
in developmental language disorders.
Nonetheless, it appears that the
quality and quantity of linguistic input
in the early years impacts on childrens
linguistic achievements. Hart and Risley
[1995] discussed the importance of the
cumulative effect of linguistic input
over time, with children with more
enriched input having larger vocabula-
ries at every time point. The main
factor associated with the level of lin-
guistic input in this study was socio-
economic status (SES). Subsequent
authors [e.g., Hoff, 2003] proposed that
variation in linguistic input was medi-
ated by SES; however two recent large-
scale population studies found little rela-
tionship between SES and LLE
[Zubrick et al., 2007; Reilly et al.,
2009]. Conversely, SES has been associ-
ated with SLI in 46 year old children
[Stanton-Chapman et al., 2002; Reilly
et al., 2010]. However, Plomin et al.
[1997] found that children adopted at
birth became increasingly similar to
their biological parents in cognitive and
verbal abilities from infancy through
adolescence than to their adoptive
parents, confirming the importance of
genetic influences on language abilities.
It appears that associations between lin-
DEV DISABIL RES REV  EARLY LANGUAGE DELAY
guistic environment, SES, and language
ability may be mediated by genetic
influences.
There seems to be a differential
effect of genes and environment on lan-
guage development toward the extreme
low ends of the spectrum in environ-
ment and ability. It is undisputed that
environmental deprivation at the
extreme end of the spectrum (poverty
or severe neglect) is linked with
depressed cognitive and language devel-
opment [Turkheimer et al., 2003]. In a
large twin study, Dale et al. [1998]
reported that the genetic contribution
accounted for 73% of the variance in
vocabulary development for children in
the lowest 5% of language ability at 2
years. For the entire sample, 25% of the
variance was accounted for by genetic
variables rather than environmental
factors.
Unfolding Neurobiology
Having seen that genetic factors
play a large role in late talking and SLI
outcomes, we turn now to considering
the neurobiology of language. Signifi-
cant differences in the neurobiology of
children with SLI have been demon-
strated. For recent reviews, see Webster
and Shevell [2011] and Badcock et al.
[2012]. Brain structures in children
with SLI have been found to differ in
volume and symmetry from those of
TD children. Cohen et al. [1989] also
found differences in gyrification. Vol-
ume differences, relative to TD peers,
include a smaller Brocas area [Gauger
et al., 1997], reduced surface area in the
primary auditory cortex of the left
hemisphere [Leonard et al., 2002], and
increased gray matter in the right peri-
sylvian region and the occipital petalia
[Soriano-Mas et al., 2009]. Typically
the dominant, left hemisphere language
areas are larger in volume compared
with the corresponding regions of the
right hemisphere. However, for chil-
dren with SLI the hemispheres may not
differ in volume, or may show a right-
ward dominance. Reduced asymmetry
of the perisylvian structures has been
found [Plante et al., 1991], particularly
in the temporal [Ors et al., 2005] and
planum temporale regions [Gauger
et al., 1997; Leonard et al., 2002]. At
odds with the latter finding is one
report of normal leftright asymmetry
in children with SLI [Preis et al., 1998].
These volumetric (and possibly
asymmetric) differences become impor-
tant in their relationship with functional
language organization and performance.
However, as Badcock et al. [2012] point
AND SLI  MOYLE ET AL.
out, there are few studies of how these
physiological differences relate to lan-
guage function, and prior to Badcock
et al., they were limited to investigations
of the KE family [e.g., Vargha-Khadem
et al., 2005]. Nonetheless, these studies
do point to areas of the brain where
neurolinguistic processing may be differ-
ent for children with SLI, and
potentially, children who are LTs.
COGNITIVE MECHANISMS
Perception
Measures of perception have been
found to correlate with language devel-
opment. Speech perception plays an
important role in language acquisition.
Speech perception skills, including pho-
neme discrimination and learning to
inhibit processing of non-critical sound
contrasts, in infancy have been found to
correlate with later language develop-
ment [Conboy et al., 2008]. The
assumption is that the sooner infants are
able to discriminate between their
native language phonemes, the sooner
they can map phonological patterns and
begin to learn words. Behavioral and
electrophysiological measurements, such
as event related potentials (ERP), have
been used to show that infants native
language speech discrimination abilities
measured at seven months predict lan-
guage outcomes at 1430 months.
Those with better native speech dis-
crimination had significantly higher
language abilities whereas those with
better non-native phonetic discrimina-
tion had poorer language abilities (see
Kuhl, [2010], for a review). Children
who progress more quickly toward
neural commitment to native sounds
contrasts progress faster than those who
retain the ability to discriminate
between non-native contrasts for longer
[Kuhl, 2010]. Children who advance in
phonetic abilities faster should begin to
detect phonotactic patterns and words
more readily, leading to faster develop-
ment of vocabulary and subsequent
grammar. In a series of ERP studies,
Molfese et al. found that infants who
could discriminate consonants better
between phonemic boundaries had bet-
ter language development at ages 3, 5,
and 9 years old [Molfese et al., 1999].
In addition, Tsao et al. [2004] found
that infants who could discriminate
between two non-native vowels at six
months of age had better language
development at 13, 16, and 24 months.
Newman et al. [2006] found that in
TD infants, the ability to detect a word
in a stream of connected speech
163
correlated with language outcomes at 2
and at 46 years of age. This ability was
not related to general intelligence meas-
ures, but correlated with vocabulary
and grammar.
Speed of Processing
Given the rapid transient nature
of the acoustic speech signal, it is logical
that efficient processing of brief audi-
tory temporal stimuli may be an
important cognitive process underlying
the development of language. Difficul-
ties with rapid auditory temporal
processing have been associated in some
children with SLI [Tallal et al., 1985;
Bishop et al., 1999], and may poten-
tially play a contributing role in the
disorder. Studies have found correla-
tions between rapid temporal processing
of acoustic signals and early language
development. For example, Benasich
et al. [2008] found that children with a
family history of language disorders
were not as efficient at detecting very
rapid stimuli (70 ms between stimuli) as
those without a family history. The dif-
ferences in latency of response to these
rapid signals at six months of age pre-
dicted language outcomes at 24
months. These correlations have been
found to hold through ages 3 and 4 for
cognitive and language measures
[Choudhury and Benasich, 2011].
The speed of spoken word recog-
nition has also been investigated as a
factor in language development. Fernald
et al. [2006] demonstrated that efficient
processing of spoken language correlates
with concurrent and later vocabulary
development in TD infants and tod-
dlers. In addition, speed of spoken
word recognition at 25 months of age
contributes unique variance to language
and cognitive skills at age 8 years
[Marchman and Fernald, 2008]. Fernald
and Marchman [2012] investigated the
processing speed of LTs in relation to
their language outcomes 18 months
later. They found those who started out
in the lowest quintile at 18 months of
age, but had faster processing speeds,
were more likely to show accelerated
vocabulary growth over the next 18
months, compared with LTs with less
efficient processing. Their data are con-
sistent with a model of cascading
influences of processing on later devel-
opment. This is a promising factor to
consider in the prediction of language
disorders in LTs.
Working Memory
A complete review of working
memory models is beyond the scope of
164
this review (see Baddeley [2012] and
Courage and Cowan [2009], for over-
views). Here we briefly review why
working memory skills are relevant for
language learning, adopting Baddeleys
model. In this account, working mem-
ory, a cognitive system that mediates
between perception, long-term mem-
ory, and action [Baddeley, 2012, p.
25] is considered to be a four-compo-
nent cognitive system comprised of a
central executive, an episodic buffer, a
phonological loop, and a visuo-spatial
sketchpad. All but the latter are relevant
here. The central executive is an atten-
tion system, controlling directed,
focused, and sustained attention, and
attention shifts, while inhibiting irrele-
vant information. The episodic buffer
acts as a temporary store for chunks of
information that link working memory,
perception, and long-term memory.
The phonological loop is believed to be
necessary for learning new word-form
to referent mappings (learning new
words). This sub-system acts as a tem-
porary store of newly encountered
phonological forms that are linked to
long-term memory with increased ex-
perience. The loop is susceptible to
rapid decay and capacity limitations,
both of which improve with age and
language experience.
Executive functioning and lan-
guage usually show moderate
correlations in the 36 year age range;
however this is thought to be due to the
role language plays in helping children
regulate their attention and behavior
purposefully, rather than vice versa
[Carlson, 2005; Hughes and Ensor,
2007]. Working memory skills are
strongly linked to vocabulary learning in
preschool children and deficits in work-
ing memory have been identified in
children with SLI [e.g., Gathercole and
Baddeley, 1990]. Deficits in complex
working memory, inhibition, and atten-
tion skills are associated with SLI at a
group level throughout early-to-middle
childhood [Alloway and Gathercole,
2006; Im-Bolter et al., 2006; Marton,
2008].
The use of working memory as a
predictor factor is in its early stages of
investigation. However, there is some
research on this related to predicting
language outcomes in young children.
Conboy et al. [2008] used behavioral
measures such as retrieving objects hid-
den or placed in boxes to estimate
development of executive function.
They found that in 11-month-old
infants, scores on these tasks were cor-
related with the ability to ignore speech
DEV DISABIL RES REV  EARLY LANGUAGE DELAY
variations that were irrelevant to their
native language. Infants native speech
discrimination was also related to their
concurrent receptive vocabulary skills.
The authors suggested the role of inhi-
bition in development of neural
commitment to native speech contrasts.
Studies have measured neural ac-
tivity in the frontal regions associated
with cognitive processes such as atten-
tion, working memory, and associative
learning in relation to predicting lan-
guage. This brain activity, known as
resting frontal gamma power, at 16, 24,
and 36 months of age significantly cor-
related with sentence structure scores at
4 and 5 years of age [Gou et al., 2011].
In a similar study, children with a fam-
ily history of language disorders have
been found to have lower gamma
power density functions [Benasich
et al., 2008]. This research indicates
that better cognitive control and work-
ing memory skills at critical points in
development may facilitate language
development.
Phonological Short-Term Memory
Phonological short-term memory
(PSTM) is a linguistic processing com-
ponent of working memory. PSTM is
usually measured by repetition of a set
of nonsense words (nonword repetition;
NWR), increasing in complexity from
one to four or five syllables in length.
There are several published measures
available [e.g., see Gathercole et al.,
1994; Dollaghan and Campbell, 1998].
PSTM is believed to play an important
role in early language development
[Gathercole and Baddeley, 1990]. A
reduced capacity to store and process
incoming phonological information leads
to weaker or incomplete phonological
representations of words, which are then
more difficult to access and retrieve for
production, resulting in slowed vocabu-
lary acquisition [Gathercole and
Baddeley, 1990]. The ability to recall
sequences of words relies principally on
PSTM [Baddeley et al., 1998]. There is
substantial evidence to suggest that a li-
mitation in PSTM could be a factor in
language disorders [Graf Estes et al.,
2007]. Within the SLI population, few
children have PSTM abilities within the
normal range; however, when it does
occur, better PSTM ability is correlated
with higher language and literacy per-
formance [Gathercole and Baddeley,
1990; Botting and Conti-Ramsden,
2001; Alloway and Gathercole, 2006].
Recent research has indicated that
LTs also perform poorly on a test of
NWR relative to their age-matched
AND SLI  MOYLE ET AL.
peers. For example, Stokes and Klee
[2009a,b] found that NWR was
strongly related to expressive vocabulary
size in 2-year-old children. In addition,
Chiat and Roy [2008] reported that
early NWR skills at age 2 and 3 were
predictive of morphosyntactic develop-
ment at age 4 and 5. It should be noted
though that LTs with very few words
generally will not participate in a
NWR task. Stokes and Klee [2009a,b]
reported that children who did not
cooperate for NWR testing scored sig-
nificantly lower on cognitive and
language tests than children who did
cooperate. Noncompliance notwith-
standing, NWR is promising as a
predictive measure of later SLI for LTs.
THEORETICAL ACCOUNTS
The overview of the multifacto-
rial model is reflective of our
theoretical perspective of SLI and the
relationship between late talker status
and subsequent development of LI. We
adopt a domain general, rather than a
domain specific perspective. Domain
specific theories include a specific
grammatical deficit [van der Lely, 2005]
and a delay in setting the parameters of
the grammatical system [Rice et al.,
1995]. However these grammatical the-
ories are challenged by evidence of
nonlinguistic impairments. Conse-
quently, we do not discuss domain
specific theories (or linguistic theories)
in this review. Domain general theories
implicate poor temporal auditory proc-
essing [Tallal et al., 1985; Bishop et al.,
1999], reduced verbal working memory
capacity, either specific to phonological
processing [Gathercole and Baddeley,
1990; Chiat, 2001] or more broadly for
linguistic processing [Ellis Weismer and
Evans, 2002; Montgomery and Evans,
2009; Baird et al., 2010;], a generalized
slow speed of processing [Ellis Weismer
and Hesketh, 1996], and weaknesses in
the procedural memory system [Ullman
and Pierpont, 2005].
The domain general theories
emphasize a cascading effect of lower
level processing difficulties impacting
on the development of higher level
skills. Investigations of how children
with SLI learn in tasks that recruit the
procedural memory system have con-
tributed to the current focus on
learning mechanisms in children with
SLI. Ullmans [2001] declarative-proce-
dural model of language postulated two
separate (but possibly interactive) mem-
ory systems. It is postulated that the
procedural memory system, anchored in
the interactive network comprising the
DEV DISABIL RES REV  EARLY LANGUAGE DELAY
prefrontal cortex, basal ganglia, and cer-
ebellum, is responsible for the learning
and storage of rule-based behaviors,
including some motor [e.g., bike-ri-
ding] and language behaviors (e.g.,
syntax and phonology). The declarative
memory system, anchored in the medial
temporal lobe network, particularly the
hippocampus, is responsible for the
learning and storage of arbitrary (non-
rule-based) events, facts, and life experi-
ences, such as form-referent mappings
in word learning.
Ullman and Pierpont [2005] pro-
posed the Procedural Deficit
Hypothesis (PDH) as an account for
SLI, suggesting that SLI arises from a
deficit in the procedural memory sys-
tem and underlying brain structures. (In
addition to the above sections on
genetics and neurobiology, see Ullman
and Pierpont [2005] and Ullman [2001]
for an overview of the neural correlates
of SLI in relation to the PDH account.)
In this account, a grammatical LI
reflects difficulty with learning rule-
based language knowledge (syntax,
morphology, and phonology) due to
deficits in the memory system responsi-
ble for rule-based learning, the
procedural memory system.
Recent empirical evidence sup-
ports the PDH, highlighting the
difficulties that children with SLI have
with both nonverbal and verbal proce-
dural learning tasks. Tomblin et al.
[2007] reported poor learning by ado-
lescents with SLI on a visual procedural
learning task (the Serial Reaction Time
task), as did Lum et al. [2010] in their
study of 7-year-olds with SLI. How-
ever, Lum et al. [2010] also reported
significance between group differences
on a verbal declarative memory task
(the number of semantically unrelated
word pairs recalled in a list task), sug-
gesting that memory deficits may
underlie lexical learning in SLI.
As Lum and Kidd [in press]
indicated, while the procedural mem-
ory system is principally concerned
with learning information that is se-
quential and probabilistic/statistical
(e.g., grammar), and the declarative
memory system is principally con-
cerned with the association between
stimuli to store a memory (e.g., map-
ping a heard phonological form to a
referent in the visual field), the two
systems interact. Further, working
memory and the declarative and pro-
cedural memory systems also interact,
given shared neurological substrates,
particularly the prefrontal cortex [Lum
et al., in press].
AND SLI  MOYLE ET AL.
Given the observed deficits in vis-
ual procedural learning, verbal
declarative learning, and NWR, which
together are indicative of deficits in
procedural, declarative, and working
memory, Lum and Bleses [2012] sug-
gested that a primary deficit in the
latter may underlie observed deficits in
the former. They showed that the poor
performance of school-aged children
(mean age 7.6 years) with SLI on a
verbal declarative memory task was
mediated by poor working memory
abilities. Overall, there is an emerging
view that school-aged children with
SLI show memory deficits and difficul-
ties with statistical learning [e.g., Evans
et al., 2009].There is also evidence of a
strong association between implicit sta-
tistical learning and grammatical ability
in TD pre-school children [Kidd,
2012].
IMPLICIT STATISTICAL
LEARNING OF LANGUAGE IN
TODDLERS
The findings on the role of
implicit learning in SLI prompted
Stokes et al. [Stokes, 2010; Stokes
et al., 2012a, 2012b] to explore
whether implicit learning deficits or dif-
ferences could be detected in very
young children who are LTs and hence
at risk of SLI. This research focused on
probable statistical cues to language
learning during infancy. The language
input that infants hear is often a contin-
uous string of language without word
boundaries (thedogischasingthecat). This
continuous string would seem to pro-
vide a significant challenge to infants
who need to crack the code of lan-
guage learning [Kuhl, 2004]. To
overcome this challenge, infants take
advantage of statistical (probabilistic)
cues in the input that help in the iden-
tification of words [Saffran, 2003;
Saffran and Graf Estes, 2006]. These
cues include cross-syllable transitional
(sequential) probabilities of sound
sequences [Aslin et al., 1998; Thiessen
et al., 2005] and recurring clusters of
syllables [Swingley, 2005]. None of this
learning occurs with feedback and it
requires repeated or sustained input, all
of which are the hallmarks of proce-
dural learning. The question that Stokes
et al. addressed was whether there
might be statistical cues in the language
input at the lexical, rather than sub-lex-
ical, level that might influence
vocabulary acquisition in toddlers. Such
lexical cues would include phonological
neighborhood density and word
frequency.
165

Figure 2 Scatterplot of the relationship between neighborhood density and expressive vocabu-
lary size (modified from Stokes, 2010). [Color figure can be viewed in the online issue, which is
available at wileyonlinelibrary.com.]
Toddlers first spoken words are
influenced by the sound (phonological)
and whole word (lexical) characteristics
of words in the ambient language, and
the statistical regularities with which
they occur [Stokes, 2010]. For example,
toddlers first words are short words
that sound similar to many other words
(those that share phonological strings)
in the ambient language (e.g. cat has 35
similar words, or neighbors, such as
mat, pat, cap, kit). Such words have
high phonological neighborhood den-
sity (ND). Words that have few
phonological neighbors are said to
reside in sparse neighborhoods [mouth
has six neighbors: math, mouse, myth,
moth, south, mouth (verb)]. Stokes et al.
[Stokes, 2010; Stokes et al., 2012a,
2012b] showed that the relationship
between phonological ND and vocabu-
lary size was very strong in 2-year-old
children. Figure 2 shows the strength of
the relationship for English-speaking
children. The figure shows that when
vocabulary size is very small (lower
than 21.25 SDs from the mean of 0;
z-scores), childrens words are, on aver-
age, drawn from words in the input
that have very dense phonological
neighborhoods (more than 2 SDs above
the mean ND of 0; z-scores).
The finding is remarkably robust
across languages. In three languages
(English, French, and Danish), children
who were poor vocabulary learners had
expressive lexicons comprised of far
fewer sparse words than TD children of
the same age. The amount of variance
in expressive vocabulary size accounted
for by phonological ND was 47%, 53%,
166
and 41% for English, French, and Dan-
ish, respectively [Stokes, 2010; Stokes
et al., 2012a, 2012b].
Extended Statistical Learning
On the basis of these results,
Stokes et al. proposed a theory of
Extended Statistical Learning to account
for slow vocabulary growth. This
theory argues that poor learners have a
prolonged period of use of high phono-
logical density as a cue to word
learning. This initially successful statisti-
cal cue effectively blocks subsequent
learning of words from sparse neighbor-
hoods. The idea of blocking the use of
more effective statistical learning mech-
anisms had first been proposed by Aslin
and Newport [2008], although these
authors did not suggest specific mecha-
nisms, such as using ND as a cue to
learning. The question for clinicians
and researchers is why would LTs be
learning words from dense neighbor-
hoods? We believe the answer may lie
in differences in working memory abil-
ities, given that Stokes and Klee
[2009a] found that children with small
expressive vocabularies performed
poorly on a test of NWR.
Poor working memory abilities
may limit the ability to achieve form-
referent mapping of words from sparse
phonological neighborhoods because
these phonological strings occur infre-
quently in the ambient language input
[Stokes et al., 2012b]. Hsu and Bishop
[2011] suggested that deficits in work-
ing memory ability could underpin the
ability to use statistical cues in the input
in children with SLI. Longitudinal stud-
DEV DISABIL RES REV  EARLY LANGUAGE DELAY
ies are now needed to explore whether
LTs who develop SLI, when compared
with LTs who normalize, show differ-
ent patterns of lexical learning arising
from differing learning mechanisms,
such as the use of the phonological cues
inherent in word forms, and PSTM
abilities.
CLINICAL
RECOMMENDATIONS
We can draw several clinical rec-
ommendations from this literature
review. There is strong evidence to
suggest that at least three factors at 2
years of age should be considered in
clinical assessments, a family history of
late talking, being male, and being less
than 85% of predicted birth weight.
We suggest here that linguistic process-
ing skills also be assessed. Two-year-old
children with language delay should be
screened on a test of NWR, and this
should be repeated at six-month inter-
vals. We predict that children who
become late bloomers will be the chil-
dren who score higher on a test of
NWR than those LTs who will
become language impaired in the pre-
school years. This idea is speculative at
present, and awaits research. In addi-
tion, we have suggested that children
who begin as LTs and develop into
children with SLI may have quite dif-
ferent language learning mechanisms
from their TD peers. We suggest that
language clinicians explore teaching
high and low density words to children
who are identified as later talkers, in an
effort to explore childrens learning
mechanisms.
SUMMARY
At the outset of this article, we
stated that distinguishing between the
LTs who do and do not develop a lan-
guage disorder at 45 years of age poses
a diagnostic challenge. SLI is a hetero-
geneous disorder, with various
subgroups and a changing profile for
each individual across development.
Recent research on the genetic and
neurobiological influences on language
development and impairment, including
SLI, suggest that there are likely to be
genetic variants that are related to indi-
vidual differences in neurobiological
development in children that in turn
contribute to the heterogeneity of SLI.
This heterogeneity is one of the reasons
why it is difficult to identify accurate
predictors of development from late
talker status to SLI status. Attempts to
date to identify predictive factors have
at best been moderately successful. In
AND SLI  MOYLE ET AL.
this review, we suggested that it is time
to widen the search for predictive fac-
tors, so as to include measures of
language learning mechanisms. The last
decade has produced a wealth of evi-
dence on the relationships among
memory systems and language behav-
iors. Very recent work has shown that
this relationship is present even in 2-
year-old children. Children with ex-
pressive language delays have poor
short-term memory abilities and we
suggest that this memory deficit con-
tributes to the use of learning
mechanisms that differ from those of
TD 2-year-old children. We propose a
new framework for thinking about
ELD. We propose that ELD is charac-
terized not only by the quantitative
metric of number of words produced,
or the presence/absence of word com-
binations, but also by short-term
memory abilities that impact on child-
rens learning mechanisms. Improving
our ability to predict SLI outcomes
from late talker status could depend on
the careful combination of the most
promising predictors, be they genetic,
neurobiological, and/or cognitive learn-
ing mechanisms.
REFERENCES
Alloway TP, Gathercole SE, editors. 2006. Work-
ing memory and neurodevelopmental disor-
ders. New York: Psychology Press.
Aslin RN, Newport EL. 2008. What statistical
learning can and cant tell us about language
acquisition. In: Colombo J, McCardle P,
Freund, L, editors. Infant pathways to lan-
guage: methods, models and research direc-
tions. Hove: Psychology Press. p 1529.
Aslin RN, Saffran JR, Newport EL. 1998. Com-
putation of conditional probability statistics
by human infants. Psychol Sci 9:321324.
Badcock NA, Bishop DVM, Hardiman MJ, et al.
2012. Co-localisation of abnormal brain
structure and function in specific language
impairment. Brain Lang 120:310320.
Baddeley A. 2012. Working memory: theories,
models and controversies. Annu Rev Psy-
chol 63:129.
Baddeley A, Gathercole S, Papagno C. 1998. The
phonological loop as a language learning de-
vice. Psychol Rev 105:158173.
Baird G, Dworzynski K, Slonims V, et al. 2010.
Memory impairment in children with lan-
guage impairment. Dev Med Child Neurol
52:535540.
Benasich AA, Gou Z, Choudhury N, et al. 2008.
Early cognitive and language skills are linked
to resting frontal gamma power across the
first 3 years. Behav Brain Res 195:215222.
Bishop DVM. 1994. Is specific language impair-
ment a valid diagnostic category? Genetic
and psycholinguistic evidence. Philos Trans
R Soc Lond B Biol Sci 346:105111.
Bishop DVM. 2009. Genes, cognition, and com-
munication: insights from neurodevelopmen-
tal disorders. Ann NY Acad Sci 1156:118.
DEV DISABIL RES REV  EARLY LANGUAGE DELAY
Bishop DVM, Carlyon R, Deeks J, et al. 1999.
Auditory temporal processing impairment:
neither necessary nor sufficient for causing
language impairment in children. J Speech
Lang Hear Res 42:12951310.
Bishop DVM, Edmundson A. 1987. Language
impaired 4 year olds: distinguishing transient
from persistent impairment. J Speech Hear
Disord 52:156173.
Bishop DVM, Price TS, Dale PS, et al. 2003.
Outcomes of early language delay II: etiol-
ogy of transient and persistent language diffi-
culties. J Speech Lang Hear Res 46:561
575.
Bishop DVM, Snowling MJ. 2004. Developmen-
tal dyslexia and specific language impair-
ment: same or different? Psychol Bull
130:858886.
Bornstein MH, Haynes MO, Painter KM. 1998.
Sources of child vocabulary competence: a
multivariate model. J Child Lang 25:367393.
Botting N, Conti-Ramsden G. 2001. Non-word
repetition and language development in chil-
dren with specific language impairment (SLI).
Int J Lang Commun Disord 36:421432.
Buschmann A, Jooss B, Rupp A, et al. 2008. Chil-
dren with developmental language delay at 24
months of age: results of a diagnostic work-
up. Dev Med Child Neurol 50:223229.
Bricker D, Squires J. 1999. Ages & Stages Question-
naires (ASQ): a parent-completed, child-mon-
itoring system, 2nd ed. Baltimore: Brookes.
Carlson SM. 2005. Developmentally sensitive
measures of executive function in preschool
children. Dev Neuropsychol 28:595616.
Catts HW, Adolf SM, Hogan TP, et al. 2004.
Are specific language impairment and dys-
lexia distinct disorders? J Speech Lang Hear
Res 48:13781396.
Chiat S. 2001. Mapping theories of developmen-
tal language impairment: premises, predic-
tions and evidence. Lang Cogn Process
16:113142.
Chiat S, Roy P. 2008. Early phonological and
sociocognitive skills as predictors of later lan-
guage and social communication outcomes. J
Child Psychol Psychiatry 49:635645.
Choudhury N, Benasich AA. 2003. A family
aggregation study: the influence of family
history and other risk factors on language
development. J Speech Lang Hear Res
46:261272.
Choudhury N, Benasich AA. 2011. Maturation of
auditory evoked potentials from 6 to 48
months: prediction to 3 and 4 year language
and cognitive abilities. Clin Neurophysiol
122:320338.
Clegg J, Hollis C, Mawhood L, et al. 2005. De-
velopmental language disordersa follow-
up later in adult life. Cognitive, language
and psychosocial outcomes. J Child Psychol
Psychiatry 46:128149.
Cohen M, Campbell R, Yaghmai F. 1989. Neu-
ropathological abnormalities in developmen-
tal dysphasia. Ann Neurol 25:567570.
Conboy BT, Sommerville JA, Kuhl PK. 2008.
Cognitive control factors in speech perception
at 11 months. Dev Psychol 44:15051512.
Conti-Ramsden G, Botting N. 1999. Classifica-
tion of children with specific language
impairment: longitudinal considerations. J
Speech Lang Hear Res 42:11951204.
Courage ML, Cowan N, editors. 2009. The de-
velopment of memory in infancy and child-
hood. Hove, UK: Psychology Press.
Creusere M, Alt M, Plante E. 2004. Recognition
of vocal and facial cues to affect in lan-
guage-impaired and normally-developing
preschoolers. J Commun Disord 37:520.
AND SLI  MOYLE ET AL.
Dale PS, Price TS, Bishop DVM, et al. 2003. Out-
comes of early language delay. I. Predicting per-
sistent and transient language difficulties at 3 and
4 years. J Speech Lang Hear Res 46:544560.
Dale PS, Simonoff E, Bishop DVM, et al. 1998.
Genetic influence on language delay in two-
year-old children. Nat Neurosci 1:324328.
Desmarais C, Sylvestre A, Meyer F, et al. 2008.
Systematic review of the literature on char-
acteristics of late-talking toddlers. Int J Lang
Commun Disord 43:361389.
Dollaghan CA, Campbell TF. 1998. Nonword
repetition and child language impairment. J
Speech Lang Hear Res 41:11361146.
Dollaghan CA, Campbell TF, Paradise JL, et al.
1999. Maternal education and measures of
early speech and language. J Speech Lang
Hear Res 42:14321443.
Ellis E, Thal D. 2008. Early language delay and
risk for language impairment. Perspect Lang
Learn Educ 15:93100.
Ellis Weismer SE, Evans JL. 2002. The role of
processing limitations in early identification
of specific language impairment. Top Lang
Disord 22:1529.
Ellis Weismer SE, Hesketh LJ. 1996. Lexical
learning by children with specific language
impairment: effects of linguistic input pre-
sented at varying speaking rates. J Speech
Hear Res 39:177190.
Evans JL, Saffran JR, Robe-Torres K. 2009.
Statistical learning in children with specific
language impairment. J Speech Lang Hear
Res 52:321335.
Fenson L, Marchman VA, Thal D, et al. 2007.
The MacArthur-Bates Communicative De-
velopment Inventories: users guide and tech-
nical manual, 2nd ed. Baltimore: Brookes.
Fernald A, Marchman VA. 2012. Individual dif-
ferences in lexical processing at 18 months
predict vocabulary growth in typically
developing and late-talking toddlers. Child
Dev 83:203222.
Fernald A, Perfors A, Marchman VA. 2006. Pick-
ing up speed in understanding: speech proc-
essing efficiency and vocabulary growth
across the 2nd year. Dev Psychol 42:98116.
Finneran DA, Francis AL, Leonard LB. 2009.
Sustained attention in children with specific
language impairment (SLI). J Speech Lang
Hear Res 52:915929.
Fisher SE, Scharff C. 2009. FOXP2 as a molecu-
lar window into speech and language.
Trends Genet 25:166177.
Flax JF, Realpe-Bonilla T, Roesler C, et al. 2009.
Using early standardized language measures
to predict later language and early reading
outcomes in children at high risk for lan-
guage-learning impairments. J Learn Disabil
42:6175.
Gathercole SE, Baddeley AD. 1990. Phonological
memory deficits in language disordered chil-
dren: is there a causal connection? J Mem
Lang 29:336360.
Gathercole SE, Willis CS, Baddeley AD, et al.
1994. The children test of nonword repeti-
tion: a test of phonological working mem-
ory. Memory 2:103127.
Gauger LM, Lombardino LJ, Leonard CM. 1997.
Brain morphology in children with specific
language impairment. J Speech Lang Hear
Res 40:12721284.
Gottlieb G. 2007. Probabalistic epigenesis. Dev
Sci 10:111.
Gou Z, Choudhury N, Benasich AA. 2011. Rest-
ing frontal gamma power at 16, 24 and 36
months predicts individual differences in
language and cognition at 4 and 5 years.
Behav Brain Res 220:263270.
167
Graf Estes K, Evans JL, Else-Quest NM. 2007.
Differences in nonword repetition perform-
ance of children with and without specific
language impairment: a meta-analysis. J
Speech Lang Hear Res 50:177195.
Hart B, Risley TR. 1995. Meaningful differences
in the everyday experiences of young Amer-
ican children. Baltimore, MD: Paul H.
Brookes.
Hill EL. 2001. Non-specific nature of specific
language impairment: a review of the litera-
ture with regard to concomitant motor
impairments. Int J Lang Commun Disord
36:149171.
Hoff E. 2003. The specificity of environmental
influence: socioeconomic status affects early
vocabulary development via maternal
speech. Child Dev 74:13681378.
Hughes C, Ensor R. 2007. Executive function and
theory of mind: predictive relations from ages
2 to 4. Dev Psychol 43:14471459.
Hsu HJ, Bishop DVM. 2011. Grammatical diffi-
culties in children with specific language
impairment: is learning deficient? Hum Dev
53:264277.
Im-Bolter N, Johnson J, Pascual-Leone J. 2006.
Processing limitations in children with spe-
cific language impairment: the role of exec-
utive function. Child Dev 77:18221841.
Johnston JR. 1994. Cognitive abilities of children
with language impairment. In: Watkins RV,
Rice ML, editors. Specific language impair-
ments in children. Baltimore, MD: Paul H.
Brookes. p 107120.
Kamhi AG, Catts HW, Koenig LA, et al. 1984.
Hypothesis-testing and nonlinguistic sym-
bolic abilities in language-impaired children.
J Speech Hear Disord 49:169176.
Kidd E. 2012. Implicit statistical learning is
directly associated with the acquisition of
syntax. Dev Psychol 48:171184.
Klee T, Carson DK, Gavin WJ, et al. 1998. Con-
current and predictive validity of an early
language screening program. J Speech Lang
Hear Res 41:627641.
Klee T, Pearce K, Carson DK. 2000. Improving
the positive predictive value of screening for
developmental language disorder. J Speech
Lang Hear Res 43:821833.
Klee T, Stokes SF. 2011. Language development.
In: Skuse D, Bruce H, Dowdney L, Mrazek
D, editors. Child psychology and psychiatry:
frameworks for practice, 2nd ed. Chichester:
Wiley-Blackwell. p 4550.
Kuhl PK. 2004. Early language acquisition: crack-
ing the speech code. Nat Rev Neurosci
5:831843.
Kuhl PK. 2010. Brain mechanisms in early lan-
guage acquisition. Neuron 67:713727.
Leonard CM, Lombardino LJ, Walsh K, Eckert
MA, Mockler JL, Rowe LA, Williams S,
DeBose CB. 2002. Anatomical risk factors
that distinguish dyslexia from SLI predict
reading skill in normal children. Jnl Comm
Dis 35:501531.
Leonard LB. 1998. Children with specific lan-
guage impairment. Cambridge, MA: MIT
Press.
Leonard LB, Ellis Weismer SE, Miller CA, et al.
2007. Speed of processing, working mem-
ory, and language impairment in children. J
Speech Lang Hear Res 50:408428.
Lum JAG, Bleses D. 2012. Declarative and proce-
dural memory in Danish speaking children
with specific language impairment. J Com-
mun Disord 45:4658.
168
Lum JAG, Conti-Ramsden G, Page D, et al.
2012. Working, declarative, and procedural
learning in specific language impairment.
Cortex 48:11381154.
Lum JAG, Gelgic C, Conti-Ramsden G. 2010.
Procedural and declarative memory in chil-
dren with and without specific language
impairment. Int J Lang Commun Disord
45:96107.
Lum JAG, Kidd E. 2012. An examination into
the associations between multiple memory
systems, past tense and vocabulary in typi-
cally developing 5-year old children. J
Speech Lang Hear Res 55:9891006.
Marchman VA, Fernald A. 2008. Speed of word
recognition and vocabulary knowledge in
infancy predict cognitive and language out-
comes in later childhood. Dev Sci 11:916.
Marton K. 2008. Visuo-spatial processing and ex-
ecutive functions in children with specific
language impairment. Int J Lang Commun
Disord 43:181200.
Molfese DL, Molfese VJ, Espy KA. 1999. The
predictive use of event-related potentials in
language development and the treatment of
language disorders. Dev Neuropsychol
16:373377.
Montgomery JW. 1993. Haptic recognition of
children with specific language impairment:
effects of response modality. J Speech Hear
Res 36:98104.
Montgomery JW, Evans JL. 2009. Complex sen-
tence comprehension and working memory
in children with specific language impair-
ment. J Speech Lang Hear Res 52:269288.
Montgomery JW, Magimairaj BL, Finney MC.
2010. Working memory and specific lan-
guage impairment: an update on the relation
and perspectives on assessment and treat-
ment. Am J Speech Lang Pathol 19:7894.
Newman R, Ratner NB, Jusczyk AM, et al.
2006. Infants early ability to segment the
conversational speech signal predicts later
language development: a retrospective analy-
sis. Dev Psychol 42:643655.
Olswang LB, Rodriguez B, Timler G. 1998.
Recommending intervention for toddlers
with specific language learning difficulties.
Am J Speech Lang Pathol 7:2332.
Oram Cardy JE, Flagg EJ, Roberts W, et al.
2008. Auditory evoked fields predict lan-
guage ability and impairment in children.
Int J Psychophysiol 68:170175.
Ors M, Ryding E, Lindgren M, et al. 2005.
SPECT findings in children with specific
language impairment. Cortex 41:316326.
Paul R. 1996. Clinical implications of the natural
history of slow expressive language develop-
ment. Am J Speech Lang Pathol 5:521.
Paul R, Elwood TJ. 1991. Maternal linguistic
input to toddlers with slow expressive lan-
guage development. J Speech Hear Res
34:982988.
Plante E, Swisher L, Vance R, et al. 1991. MRI
findings in boys with specific language
impairment. Brain Lang 41:5266.
Plomin R, Fulker DW, Corley R, et al. 1997.
Nature, nurture, and cognitive development
from 1 to 16 Years: a parent-offspring adop-
tion study. Psychol Sci 8:442447.
Preis S, Jancke L, Schittler P, et al. 1998. Normal
intrasylvian anatomical asymmetry in chil-
dren with developmental language disorder.
Neuropsychologia 36:849855.
Price TS, Eley TC, Dale PS, et al. 2003. Genetic
and environmental covariation between
verbal and nonverbal cognitive development
in infancy. Child Dev 71:948959.
DEV DISABIL RES REV  EARLY LANGUAGE DELAY
Reilly S, Bavin EL, Bretherton L, et al. 2009.
The Early Language in Victoria Study
(ELVS): a prospective, longitudinal study of
communication skills and expressive vocabu-
lary development at 8, 12 and 24 months.
Int J Speech Lang Pathol 11:344357.
Reilly S, Wake M, Bavin EL, Prior M, Williams
J, Bretherton L et al. 2007. Predicting lan-
guage at 2 years of age: A prospective com-
munity study. Pediatrics 120:14411449.
Reilly S, Wake M, Ukoumunne OC, et al. 2010.
Predicting language outcomes at four years
of age: findings from early language in Vic-
toria study. Pediatrics 126:15301537.
Rescorla L. 1989. The Language Development Sur-
vey: a screening tool for delayed language in
toddlers. J Speech Lang Hear Res 54:587599.
Rescorla L, Roberts J. 1997. Late talkers at 2: out-
come at age 3. J Speech Lang Hear Res
40:556566.
Rice ML, Taylor CL, Zubrick SR. 2008. Lan-
guage outcomes of 7-year-old children with
or without a history of late language emer-
gence at 24 months. J Speech Lang Hear
Res 51:394407.
Rice ML, Tomblin JB, Hoffman L, et al. 2004.
Grammatical tense deficits in children with
SLI and nonspecific language impairment:
relationships with nonverbal IQ over time. J
Speech Lang Hear Res 47:816834.
Rice ML, Wexler K, Cleave PL. 1995. Specific
language impairment as a period of extended
optional infinitive. J Speech Hear Res
38:850863.
Rowe ML, Raudenbush SW, Goldin-Meadow S.
2011. The pace of vocabulary growth helps
predict later vocabulary skill. Child Devt
83: 508525.
Saffran JR. 2003. Statistical language learning:
mechanisms and constraints. Curr Dir Psy-
chol Sci 12:110114.
Saffran JR, Graf Estes K. 2006. Mapping sound to
meaning: connections between learning
about sounds and learning about words. Adv
Child Dev Behav 34:138.
Soriano-Mas C, Pujol J, Ortiz H, et al. 2009.
Age-related brain structural alterations in
children with specific language impairment.
Hum Brain Mapp 30:16261636.
Stanton-Chapman TL, Chapman DA, Bainbridge
NL, et al. 2002. Identification of early risk
factors for language impairment. Res Dev
Disabil 23:390405.
Stokes SF. 1997. Secondary prevention of paedi-
atric language disability: a comparison of
parents and nurses as screening agents. Int J
Lang Commun Disord 32:139158.
Stokes SF. 2010. Neighborhood density and word
frequency predict vocabulary size in tod-
dlers. J Speech Lang Hear Res 53:670683.
Stokes SF, Bleses D, Basbll H, et al. 2012a Sta-
tistical learning in emerging lexicons: The
case of Danish. J Speech Lang Hear Res
55:12651273.
Stokes SF, Kern S, dosSantos C. 2012b. Extended
statistical learning as an account for slow vo-
cabulary growth. J Child Lang 39:105129.
Stokes SF, Klee T. 2009a. Factors that influence
vocabulary development in two-year-old chil-
dren. J Child Psychol Psychiatry 50:498505.
Stokes SF, Klee T. 2009b. The diagnostic accu-
racy of a new test of early non word repeti-
tion for differentiating late talking and typi-
cally developing children. J Speech Lang
Hear Res 52:872882.
Stothard SE, Snowling MJ, Bishop DVM, et al.
1998. Language-impaired preschoolers: a
follow-up into adolescence. J Speech Lang
Hear Res 41:407418.
AND SLI  MOYLE ET AL.
Swingley D. 2005. Statistical clustering and the
contents of the infant vocabulary. Cogn Psy-
chol 50:86132.
Tallal P, Stark RE, Mellits ED. 1985. Identifica-
tion of language-impaired children on the
basis of rapid perception and production
skills. Brain Lang 26:314322.
Thal D. 2000. Late-talking toddlers: are they at
risk? San Diego, CA: San Diego State Uni-
versity Press.
Thal D, Tobias S. 1992. Communicative gestures in
children with delayed onset of oral expressive
vocabulary. J Speech Hear Res 35:12811289.
Thal D, Tobias S, Morrison D. 1991. Language
and gesture in late talkers. J Speech Lang
Hear Res 34:604612.
Thiessen ED, Hill EA, Saffran JR. 2005. Infant-
directed speech facilitates word segmenta-
tion. Infancy 7:5371.
Tomblin JB, Mainela-Arnold E, Zhang X. 2007.
Procedural learning and adolescents with
and without specific language impairment.
Lang Learn Dev 3:269293.
Tomblin JB, Zhang X, Weiss A, et al. 2004.
Dimensions of individual differences in
communication skills among primary grade
children. In: Rice ML, Warren SF, editors.
DEV DISABIL RES REV  EARLY LANGUAGE DELAY
Developmental language disorders: from
phenotypes to etiologies. Mahwah, NJ:
Lawrence Erlbaum. p 5389.
Tsao F, Liu H, Kuhl PK. 2004. Speech percep-
tion in infancy predicts language develop-
ment in the second year of life: a longitudi-
nal study. Child Dev 75:10651084.
Turkheimer E, Haley A, Waldron M, et al. 2003.
Socioeconomic status modifies heritability of
IQ in young children. Psychol Sci 14:623628.
Ullman MT. 2001. A neurocognitive perspective
on language: the declarative/procedural
model. Nat Rev Neurosci 2:717726.
Ullman MT, Pierpont EI. 2005. Specific language
impairment is not specific to language: the pro-
cedural deficit hypothesis. Cortex 41:399433.
van der Lely HJK. 2005. Domain-specific cogni-
tive systems: insight from Grammatical-SLI.
Trends Cogn Sci 9:5359.
Vargha-Khadem F, Gadian DG, Copp A, et al.
2005. FOXP2 and the neuroanatomy of
speech and language. Nat Rev Neurosci
6:131138.
Vargha-Khadem F, Watkins KE, Price CJ, et al.
1998. Neural basis of an inherited speech
and language disorders. Proc Natl Acad Sci
USA 95:12695700.
AND SLI  MOYLE ET AL.
Vernes SC, Newbury DF, Abrahams BS, et al.
2008. A functional genetic link between dis-
tinct developmental language disorders. N
Engl J Med 359:23372345.
Vig S, Kaminer RK, Jedrysek E. 1987. A later
look at borderline and mildly retarded pre-
schoolers. J Dev Behav Pediatr 8:1217.
Vigil DC, Hodges J, Klee T. 2005. Quantity and
quality of parental language input to late-
talking toddlers during play. Child Lang
Teach Ther 21:107122.
Webster RI, Shevell MI. 2011. Topical review:
neurobiology of specific language impair-
ment. J Child Neurol 19:471481.
Whitehouse AJO, Bishop DVM, Ang QW,
et al. 2011. CNTNAP2 variants affect
early language development in the general
population. Genes Brain Behav 10:
451456.
Yoder PJ, Warren SF. 2004. Early predictors of
language in children with and without
Down syndrome. Am J Ment Retard
109:285300.
Zubrick SR, Taylor CL, Rice ML, et al. 2007.
Late language emergence at 24 months: an
epidemiological study of prevalence, predic-
tors, and covariates. J Speech Lang Hear
Res 50:15621592.
169