Sei sulla pagina 1di 14

LWW/IYC IYC200089 February 23, 2012 1:31

Infants & Young Children


Vol. 25, No. 2, pp. 158–171
Copyright C 2012 Wolters Kluwer Health | Lippincott Williams & Wilkins

Factors Associated With


Expressive and Receptive
Language in French-Speaking
Toddlers Clinically Diagnosed
With Language Delay
Audette Sylvestre, PhD; Chantal Desmarais, PhD;
François Meyer, PhD; Isabelle Bairati, PhD;
Nancie Rouleau, PhD; Chantal Mérette, PhD
The purpose of this exploratory study was to examine child and environmental factors known to
be associated to language development and how they relate to results in expressive vocabulary,
expressive language, and receptive language in language-delayed toddlers. The cross-sectional
data on 96 French-speaking children aged 18–36 months were gathered at the point of entry
into a longitudinal study of 2-year-old children displaying language delay. Measures of language,
child development, and child and environmental factors were administered. When several factors
individually associated with language development were considered concurrently, cognitive devel-
opment was consistently associated with the outcomes. Other child factors, such as male gender
and age, were also retained in the regression model explaining expressive vocabulary, whereas
only age was added in the model explaining expressive language. Two environmental factors
were involved in receptive language; that is, parental education and parental stress accounted for
8% of the variance. Factors linked to development varied across language modalities such that
parental education and parental stress were related to comprehension but not to production. The
findings suggest a strong biological trajectory for expressive language development and vocabu-
lary production, which are not affected by environmental factors. Key words: environmental
factors, expressive language, French language, language delay, receptive language, toddlers,
vocabulary

Author Affiliations: Département de réadaptation, The authors thank Lyne Champoux, Karine Messier,
Programme de maı̂trise en orthophonie, Université Hélène Crépeau, Olga Gordynska, Isabel Moreau, and
Laval, Centre Interdisciplinaire de Recherche en numerous research assistants as well as statistics con-
Réadaptation et Intégration Sociale, Québec, Canada sultants for their assistance. They also thank the SLPs
(Drs Sylvestre and Desmarais); Département de who collaborated in the recruitment of participants
médecine sociale et préventive, Université Laval, and the children and parents who generously gave of
Centre de Recherche L’Hôtel-Dieu-de-Québec, Québec, their time to participate in this study.
Canada (Dr Meyer); Département de chirurgie,
The authors have disclosed that they have no signif-
Université Laval and Direction de Santé publique de
icant relationships with, or financial interest in, any
la Capitale-Nationale, Québec, Canada (Dr Bairati);
commercial companies pertaining to this article.
École de psychologie (Dr Rouleau) and Département
de psychiatrie (Dr Merette), Université Laval, Centre Correspondence: Audette Sylvestre, PhD, Département
de Recherche Université Laval—Robert Giffard, de réadaptation, Programme de maı̂trise en ortho-
Québec, Canada. phonie, Pavillon Ferdinand-Vandry, Université Laval,
1050, avenue de la Médecine, Québec, QC, Canada G1V
This study was supported by Grant 6472 from the Fonds
0A6 (audette.sylvestre@rea.ulaval.ca).
de la recherche en santé du Québec (Quebec Fund for
Health Research). DOI: 10.1097/IYC.0b013e31823dca22

158

Copyright © 2012 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
LWW/IYC IYC200089 February 23, 2012 1:31

Factors Associated With Language Delay 159

L ANGUAGE DELAY (LD) represents the


most prevalent developmental problem
among young children (Law, 1992). Various
2008). Few researchers have addressed the
study of factors likely to be related to language
expression in a broad sense, even if expres-
studies have established that 10%–19% of in- sive language integrates a number of linguis-
fants younger than 3 years are affected by tic dimensions other than vocabulary itself,
LD without any identified underlying explana- including grammar and pragmatic skills. This
tory cause (Dale, Price, Bishop, & Plomin, is also true for factors linked to language com-
2003; Horwitz et al., 2003; Klee et al., 1998; prehension, despite the fact that a number
Reilly et al., 2007; Rescorla, 1989; Zubrick, of children younger than 3 years with lexical
Taylor, Rice, & Slegers, 2007). Having early delays often exhibit more global language dif-
LD places a child at risk to experience be- ficulties (Bishop, Price, Dale, & Plomin, 2003;
havioral and psychosocial adjustment prob- Desmarais, Sylvestre, Meyer, Bairati, &
lems at preschool age, as well as later learning Rouleau, 2010). Receptive language difficul-
difficulties (Carson, Klee, Perry, Donaghy, & ties, such as the inability to understand ques-
Muskina, 1997; Justice, Invernizzi, & Meier, tions or instructions, are much less under-
2002; Vigil, Hodges, & Klee, 2005). Language stood because children with this problem are
development depends on complex interac- frequently excluded from the studies (Hadley
tions between neurobiological and environ- & Holt, 2006; Rescorla & Merrin, 1998).
mental stimuli (Elman et al., 1996; Garbarino Within this context, the purpose of this
& Ganzel, 2000; Shonkoff & Marshall, 2000). research was to examine the associations
Therefore, to better define the targets of pre- between potential child and environmental
vention and early intervention programs for factors and early LD in terms of how they ex-
children affected by LD, one must identify the plain the developmental outcomes of 3 differ-
factors linked to this condition. ent linguistic dimensions, that is, expressive
Several studies have been carried out to in- vocabulary, expressive language, and recep-
ventory factors linked to early language de- tive language, among French-speaking Que-
velopment, and those results are summarized bec children who were clinically diagnosed
in various systematic reviews (Desmarais, with LD between the ages of 18 months and
Sylvestre, Meyer, Bairati, & Rouleau, 2008; 36 months.
Nelson, Nygren, Walker, & Panoscha, 2006;
Prathanee et al., 2009; Prathanee, Thinkham- METHODS
rop, & Dechongkit, 2007). Results reveal the
diversity of factors associated with negative This study was conducted by using en-
outcomes in language development, such as try data of a longitudinal study, including
male gender, prematurity, low birth weight, 2-year-old white French-speaking Quebec
family history of language or learning disabili- children with LD. Children were recruited be-
ties, single parenthood, small family size, late tween February 2005 and January 2007 via 10
birth order, young age of mother (<18 years) speech–language pathologists (SLPs) working
at birth of child, mother with severe depres- in health units of the Quebec City region in
sion, high parental stress, low parental educa- Canada. To participate in the study, children
tion, and low socioeconomic status. A close had to be between 18 months and 36 months
association between language and cognition old and diagnosed with LD based on the clini-
has also been reported in explanatory models cal evaluation performed by the SLPs. To carry
of language development (Bates, Tomasello, out the evaluation, the SLP met the parent and
& Slobin, 2005; Hirsch-Pasek, Golinkoff, child for one session during which a parent-
Hennon, & Maguire, 2004). child play interaction was observed and the
In studies examining risk factors of early child’s utterances were recorded. The clini-
LD, the outcome was almost exclusively ex- cian then interviewed the parent to complete
pressive vocabulary, that is, to say the num- the data collection necessary to confirm the
ber of words the child uses (Desmarais et al., presence of LD. Children were not eligible if

Copyright © 2012 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
LWW/IYC IYC200089 February 23, 2012 1:31

160 INFANTS & YOUNG CHILDREN/APRIL–JUNE 2012

their LD was associated with a known underly- those that the child uses. The total number of
ing cause. This research was approved by the words instead of the percentile rank was se-
Laval University research ethics committee. lected for the analyses because some children
in the sample were older than 30 months.
Subjects Expressive and receptive languages were
For the purpose of the current study, six evaluated by using the parental French–
participants were excluded from the entire Canadian version of the Rossetti Infant Tod-
cohort of 102 children since they spoke not dler Language Scale (ITLS; Rossetti, 1990;
only French but also a second language at an Sylvestre & St-Cyr Tribble, 2001). It is a
advanced level, relative to their age. A total criterion-referenced tool where each scale is
of 96 children (69 boys and 27 girls) whose divided into twelve 3-month intervals. It is
average ages were 29.4 months (SD = 4.4) used to compare a child’s performance to gen-
participated in the study. eral developmental standards and does not
provide a standardized score. For the pur-
Materials and procedures pose of the study herein, a score was derived
Data collection was carried out by two by using a statistical saturation procedure, by
trained research assistants during a 2-hr in- way of establishing the percentage of correct
home visit, 2–3 weeks after the SLP assess- responses related to four age intervals per
ment. One research assistant verified that scale, that is, the interval corresponding to
the parent had properly filled out the previ- the child’s age as well as to the three preced-
ously mailed French–Canadian version of the ing intervals (Desmarais et al., 2010).
MacArthur Bates Communicative Develop-
ment Inventories (MBCDI; Fenson et al., 1993; Selected factors
Trudeau, Frank, & Poulin-Dubois, 1999) used A structured questionnaire was adminis-
to evaluate the child’s expressive vocabulary. tered to evaluate the sociodemographic sta-
Additional information was collected through tus (e.g., age, education, income, and family
structured questionnaires administered face- composition); perinatal history (e.g., number
to-face with the index parent (typically of weeks of gestation, baby’s weight at birth,
the child’s mother). Those questionnaires al- and maternal postnatal depression); and his-
lowed for the gathering of information con- tory of language, hearing, or learning prob-
cerning the expressive and receptive lan- lems in the family.
guage development of the child, as well as The Parenting Stress Index—Short Form
the factors under investigation. Meanwhile, (Abidin, 1990) was designed to evaluate the
the child’s cognitive development was evalu- level of stress experienced by an adult in
ated by the second research assistant. his/her role as a parent. It consists of 36 items,
Structured questionnaires, already used in which the parent answers by using a 5-point
a large health survey and other cohort studies Likert scale. A high score reveals a high level of
conducted in the province of Quebec, were stress. The French–Canadian translation of the
administered during the home visit. original tool was validated by its authors, yield-
ing coefficients of internal coherence rang-
Language assessment ing from 0.46 to 0.82 (Bigras, LaFrenière, &
The three dependent variables measured Abidin, 1995).
were expressive vocabulary, expressive lan- The language stimulation style adopted
guage, and receptive language. Vocabulary by the parent was measured by using the
was evaluated by using the long form of the Communication Stimulation Questionnaire
16- to 30-month French–Canadian version of (Sylvestre, St-Cyr Tribble, Payette, & Cronk,
the MBCDI (Trudeau et al., 1999). It com- 1998). The tool includes seven scenarios for
prises a list of 664 words grouped in 21 cat- each of the age groups. Scenarios such as
egories among which the parent identifies “Imagine that your child has finished his meal

Copyright © 2012 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
LWW/IYC IYC200089 February 23, 2012 1:31

Factors Associated With Language Delay 161

and pushes his plate saying ‘no eat’; tell for intellectual functioning. The administra-
me what you say and do” are presented to tion was carried out in a standardized fashion
the parent. The answers are transcribed and according to the instructions provided in the
then analyzed. Twenty-seven indicators help test manual.
to pinpoint the style used for each scenario.
Language stimulation style is determined by Data analysis
adding the results of each scenario. This tool All analyses were performed using
assigns the parents’ styles to one of the two SAS/STAT software, Version 9.2 of the SAS
categories, supportive or directive (Kloth, System for Windows. Copyright 2002–2008
Janssen, Kraaimaat, & Brutten, 1998). After SAS Institute Inc. SAS and all other SAS
18 months of age, directive style is considered Institute Inc. product or service names are
to represent a child development risk factor registered trademarks or trademarks of SAS
(Hampson & Nelson, 1993). Institute Inc., Cary, NC. We first estimated the
The French–Canadian version (Kovess & bivariate association between each of the 3
Fournier, 1990) of the Composite Interna- dependent variables (expressive vocabulary,
tional Diagnostic Interview Simplified (Robins expressive language, and receptive language)
et al., 1988) was used to diagnose the pres- and the predictive factors by calculating a
ence of parental major depression during the Pearson correlation coefficient using the
last 6 months from the respondent’s point of CORR procedure of SAS or by performing
view. Analysis provides a dichotomous score, Student’s t testing for a categorical factor
indicating the presence or absence of such using the TTEST procedure of SAS. Then,
a depression. The comparison between the for each of the three outcomes, all factors
diagnoses made by using Composite Inter- associated with p ≤ .10 in the bivariate
national Diagnostic Interview Simplified with analysis were considered as candidate factors
those made by psychiatrists in both Quebec (CFs) in a multivariate linear regression
and France yields a kappa of 0.47 for depres- analysis that tests their independent and joint
sive disorders (Fournier, Lesage, Phil, Toupin, contribution on language development using
& Cyr, 1997). the REG procedure of SAS. Because of the
Finally, the Mental Development Index exploratory nature of the study, the analyses
of the Bayley Scales of Infant and Toddler were conducted by using a stepwise selection
Development–II (BSITD-II; Bayley, 1993) was procedure, specifying .05 as the significance
used to evaluate the child’s cognitive develop- level for entry and for staying in the model.
ment. This tool is designed for use with chil- For each step of the stepwise analysis, we
dren aged 1–42 months and is widely used manually calculated the degrees of freedom
in research settings with late-talking children. (df) such that the number of df for the model
The Mental Development Index scale evalu- is given by the number of CFs considered in
ates memory, habituation, problem solving, the analysis (Thompson, 1995) rather than
number concepts and generalization, classi- the number of factors retained in the model,
fication, and language and social skills. The as actually used in the “stepwise selection”
distribution of the BSITD scores is age stan- option of the REG procedure of SAS. The
dardized (within 1-month interval) to a mean number of df for the error is then n − 1 −
of 100 and a standard deviation of 15. The CF, where n is the number of subjects. This is
tool’s homogeneity is 0.88, whereas inter- explained by the fact that, at each step of the
judge agreement yields a coefficient of 0.96 stepwise selection, the factors retained in the
and a stability coefficient of 0.83. The mental model are not randomly selected but rather
scale is highly correlated with the Weschler chosen among all CFs not yet entered in
Preschool and Primary Scale of Intelligence— the model, taken into account those already
Revised (Weschler, 1989) with coefficients retained. Hence, through the entire selection
of 0.73, confirming good concurrent validity process, the number of df will be fixed to

Copyright © 2012 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
LWW/IYC IYC200089 February 23, 2012 1:31

162 INFANTS & YOUNG CHILDREN/APRIL–JUNE 2012

the number of CFs. In addition, given that (normality test), according to the results of
any stepwise selection does not guarantee to the descriptive flattening (kurtosis) and sym-
identify the best set of predictive variables, metry tests (skewness) (results not shown).
we performed another method of selection Bivariate analyses between each language
based on the coefficient of determination R2 variable and selected factors are shown in
(the RSQUARE option of the REG procedure Table 3. A review of the table shows that fac-
of SAS). Using the RSQUARE selection of SAS, tors relating to expressive vocabulary size at
we selected the best subset of independent p ≤.10 include being a boy (p = .01), child’s
variables that best predict an outcome by age (p <.0001), birth weight (p = .009), and
maximizing the R2 , fixing the subset size to cognitive development (p <.0001). In terms
the number of variables retained in the final of expressive language, being a boy (p = .03),
models of the previous stepwise selections. child’s age (p <.0001), and cognitive devel-
In the case of expressive vocabulary, the opment (p <.0001) are the factors related to
child’s age was forced into the model because development. Factors related to receptive lan-
the measurement scale does not take age guage incorporate being a boy (p = .02), ed-
into account, which is a main determinant ucation of index parent (p = .05), cognitive
of language development. This procedure development (p <.0001), number of weeks
allowed us to consider the influence of other of pregnancy (p = .05), family size (p = .07),
factors, beyond the role played by the child’s and parental stress (p = .009). Hence, the
age. The validation of each of the final models three language variables had four, three, and
was carried out by using a residual analysis six CFs, respectively, that were considered in
to verify the assumptions of homogeneity the following multivariate regressions.
of variance and normality. Noncollinearity The results from the stepwise selection pro-
and assumptions of linear regression were cedure of the multivariate regression analysis
verified on all models. The partial coefficient relating the CFs to expressive vocabulary are
of determination (partial R2 values) was used shown in Table 4. The final model, obtained in
to measure, at each step, the gain in the three steps, explains 48% of the total variance
proportion of the total variance explained by and 12% aside from the child’s age. According
the model. to this final model detailed in Table 5, boys
had fewer words in their expressive vocabu-
RESULTS lary than girls and a higher level of cognitive
development was associated with better ex-
The various sociodemographic characteris- pressive vocabulary.
tics and the distribution of the selected fac- In the multivariate regression model pre-
tors are shown in Table 1. The distribution dicting the expressive language, two factors
of the sociodemographic characteristics re- were retained, explaining 33% of the total
vealed that families were mostly in the middle variance (Table 6). Table 7 shows that both
and upper-middle socioeconomic status cate- child cognitive development (p <.0001) and
gories. age (p = .002) were highly significant. Given
Scores of language development are shown the ITLS characteristics and the fact that age
in Table 2. The MBCDI data are missing for entered the model, it seemed that being lo-
one participant. Some LD children obtained cated closer to the upper end of the bracket
low scores in one dimension of language, for each 3-month interval was associated with
for example, expressive language, and not better skills in terms of language expression.
in another, for example, receptive language, Analysis also revealed that better cognitive de-
which explains the range of scores up to velopment was associated with better expres-
100% for the ITLS. The distribution for all sive language.
three assessed dimensions is compatible with The results from the multivariate regres-
the Gaussian variable distribution hypothesis sion model created to account for all factors

Copyright © 2012 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
LWW/IYC IYC200089 February 23, 2012 1:31

Factors Associated With Language Delay 163

Table 1. Participants’ Characteristics (N = 96)

Observed
Mean (SD) Range N (%)

Child’s age (mo) 29.4 (4.4) 18–36


Child’s gender (male) 69 (71.9)
Child’s cognitive developmenta 85.8 (12.7) 50–115
Number of weeks of pregnancy 39.0 (1.6) 33.4–41.6
Birth weight (g) 3391.3 (483.9) 2060–4180
Family history of language or learning problems 13 (13.5)
Family type (other than biparental) 8 (8.3)
Family size 1.9 (0.8) 1–4
Birth order 1.6 (0.7) 1–4
Age of mothers at child’s birth 29.6 (4.6) 18–39
Postpartum depression 7 (7.3)
Major depression in past 6 monthsb 4 (4.2)
Communication stimulation style (directive)c 77 (80.2)
Parental stress (36–180)d 72.6 (16.2) 43–118
Education of respondent parent (< high school) 35 (36.5)
Income (<$40 000) 22 (22.9)

a Mental Development Index of the Bayley Scales of Infant and Toddler Development—II (Bayley, 1993).
b Composite International Diagnostic Interview Simplified (Robins et al., 1988).
c Communication Stimulation Questionnaire (Sylvestre et al., 1998).
d Parenting Stress Index—Short Form (Abidin, 1990).

simultaneously related to receptive language development had better receptive language


are shown in Table 8. Parental level of ed- (p <.0001). Children with the same level of
ucation and parental stress added 8% to the cognitive development and whose index par-
percentage of the variance already explained ents had the same level of stress had weaker
by the cognitive development, that is, 4% and scores on receptive ITLS when the parents
4%, respectively. The final model explains were less educated (p = .03). Also, after taking
31% of the variance and was obtained in three cognitive development and education level
steps. According to the final model (Table 9), into account, children whose parents had
for a given education level and parental higher stress level had lower scores on the
stress level, children with better cognitive receptive language scale (p = .005).

Table 2. Language Development Scores (N = 96)

Observed
Mean (SD) Range

Expressive vocabulary—total number of words—MBCDIa,b 163.3 (135.4) 5–552


(0–664)
Expressive language—ITLSc (0%–100%) 60.9 (24.7) 0–100
Receptive language—ITLSc (0%–100%) 78.4 (19.7) 6.2–100

aN = 95.
b MBCDI, MacArthur Bates Communicative Development Inventories (MBCDI; Fenson et al., 1993; Trudeau et al., 1999).
c ITLS, Infant Toddler Language Scale (Rossetti, 1990; Sylvestre & St-Cyr Tribble, 2001).

Copyright © 2012 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Table 3. Bivariate Analysis Between Selected Factors and Language Variables (N = 96)
LWW/IYC

Expressive
164

vocabulary Number Expressive Receptive


of words (MBCDI) Language (ITLS) Language (ITLS)
Total Sample
IYC200089

N (%) Mean (SD) p Mean (SD) p Mean (SD) p


a
Child’s gender
Male 69 (71.9) 138.7 (114.1) .01 57.4 (24.6) .03 75.9 (21.2) .02
Female 27 (28.1) 228.6 (165.4) 69.7 (23.2) 84.6 (13.7)
Family history language or learning problemsa
Yes 13 (13.5) 159.5 (127.5) .91 56.2 (28.4) .46 74.1 (23.9) .40
No 83 (86.5) 163.9 (137.3) 61.6 (24.2) 79.0 (19.0)
February 23, 2012

Family typea
Other 8 (8.3) 228.2 (170.1) .16 71.8 (25.1) .19 76.4 (8.4) .56
Biparental 88 (91.7) 157.4 (131.3) 59.9 (25.1) 78.6 (20.4)
Postpartum depressiona
1:31

Yes 7 (7.3) 140.6 (183.1) .65 52.2 (34.5) .34 72.6 (31.4) .42
No 89 (92.7) 165.1 (132.1) 61.6 (23.9) 78.8 (18.7)
Maternal major depression in past 6 monthsa
Yes 4 (4.2) 102.7 (86.7) .36 49.3 (32.4) .34 64.1 (38.6) .14
No 92 (95.8) 166.0 (136.8) 61.4 (24.4) 79.0 (18.6)
Communication stimulation style of index parenta
Directive 77 (80.2) 154.4 (134.4) .20 61.2 (23.4) .81 75.3 (18.9) .44
INFANTS & YOUNG CHILDREN/APRIL–JUNE 2012

Facilitating 19 (19.8) 199.2 (136.7) 59.6 (30.2) 75.3 (23.0)


Education of index parenta
<High school 35 (36.5) 183.9 (144.4) .26 60.5 (28.5) .92 75.6 (20.9) .05
High school and more 61 (63.5) 151.3 (129.5) 61.1 (22.5) 80.0 (18.9)
Incomea
<$40,000 22 (22.9) 179.2 (136.8) .53 59.0 (26.7) .69 73.9 (21.2) .22
$40,000 and more 74 (77.1) 158.5 (135.5) 61.4 (24.3) 79.7 (19.2)
Mean (SD) r r r
Child’s age (mo)b 29.4 (4.4) 0.60 <.0001 0.38 <.0001 − 0.01 .90
Child’s cognitive developmentb 85.8 (12.7) 0.36 <.0001 0.51 <.0001 0.48 <.0001
Number of weeks of pregnancyb 39.0 (1.6) − 0.01 .89 0.01 .91 0.20 .05
Birth weight (g)b 3391.3 (483.9) − 0.27 .009 − 0.12 .23 0.04 .71
Family sizeb 1.9 (0.8) − 0.08 .44 0.05 .60 0.19 .07
Birth orderb 1.6 (0.7) − 0.16 .12 − 0.07 .50 0.16 .13
Age of mother at child’s birthb 29.6 (4.6) − 0.15 .15 − 0.06 .57 0.02 .82
Parental stressb 72.6 (16.2) − 0.09 .38 0.02 .81 − 0.27 .009

Note. ITLS = Infant Toddler Language Scale; MBCDI = MacArthur Bates Communicative Development Inventories.
a p value resulting from a Student t test.
b p value resulting from a Pearson correlation.

Copyright © 2012 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
LWW/IYC IYC200089 February 23, 2012 1:31

Factors Associated With Language Delay 165

Table 4. Summary of Each Step of the Stepwise Procedure Based on Thompson (1995) for
Expressive Vocabulary (N = 95)

Corrected
Degree of
Candidate Freedom Sum of Squares
Factors Entered Partial
Step into the Model Model Error Model Error F p R2

1 Child’s age 4 90 614 439 1 108 065 12.5 <.0001 .36


2 Child’s age + 4 90 764 951 957 554 18.0 <.0001 .09
Child’s gender
3 Child’s age + 4 90 815 399 907 106 20.2 <.0001 .03
Child’s gender
+ Child’s
cognitive
development
Total .48

Note. For expressive vocabulary, there were four candidate factors: child’s gender; child’s age; child’s cognitive devel-
opment; and birth weight.

Table 5. Summary of Regression Analysis of Factors Related to Expressive Vocabulary


(N = 95)

95% Confidence
Nonstandardized Standardized Interval for β
Candidate Factors
Entered into the Coefficients Coefficients Lower Upper
Model β SE β β p Bound Bound

Child’s age (month) 16.97 2.40 .55 < .0001 12.20 21.74
Child’s gender (male) − 79.66 23.37 − .26 0.001 − 126.08 − 33.24
Child’s cognitive 1.89 0.84 .18 0.03 0.22 3.56
development

Table 6. Summary of Each Step of the Stepwise Procedure Based on Thompson (1995) for
Expressive Language (N = 96)

Candidate Factorsa Corrected df Sum of Squares


Entered into the
Step Model Model Error Model Error F p Partial R2

1 Child’s cognitive 3 92 15 019 43 035 10.7 <.0001 .26


development
2 Child’s cognitive 3 92 19 126 38 929 15.1 <.0001 .07
development +
Child’s age
Total .33

a For
expressive language, there were three candidate factors: child’s gender; child’s age; and child’s cognitive develop-
ment.

Copyright © 2012 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
LWW/IYC IYC200089 February 23, 2012 1:31

166 INFANTS & YOUNG CHILDREN/APRIL–JUNE 2012

Table 7. Summary of Regression Analysis of Factors Related to Expressive language (N = 96)


Nonstandardized Standardized 95% Confidence
Candidate Coefficients Coefficients Interval for β
Factors Entered
into the Model β SE β β p Lower Bound Upper Bound

Child’s cognitive .86 .17 .44 <.0001 0.52 1.20


development
Child’s age (mo) 1.54 .49 .27 .002 0.56 2.51

Table 8. Summary of Each Step of the Stepwise Procedure Based on Thompson (1995) for
Receptive Language (N = 96)

Candidate Factorsa Corrected df Sum of Squares


Entered into the
Step Model Model Error Model Error F p Partial R2

1 Child’s cognitive 6 89 8,540 28 284 4.5 .0005 .23


development
2 Child’s cognitive 6 89 10,154 26 670 5.7 <.0001 .04
development +
Parental stress
3 Child’s cognitive 6 89 11,545 25 279 6.8 <.0001 .04
development +
Parental stress +
Education of
index parent
Total .31

a For
receptive language, there were six candidate factors: child’s gender; education of index parent; child’s cognitive
development; number of week of pregnancy; family size; and parental stress.

Table 9. Summary of Regression Analysis of Factors Related to Receptive Language (N = 96)


Nonstandardized Standardized 95% Confidence
Candidate Factors Coefficients Coefficients Interval for β
Entered into the
Model β SE β β p Lower Bound Upper Bound

Child’s cognitive .71 0.13 .46 <.0001 0.45 0.98


development
Parental stress − .31 0.11 − .26 .005 − 0.53 − 0.10
Education of index − 8.16 3.63 − .20 .03 − 15.36 − 0.96
parent (<high
school)

When we performed, for each outcome, the as those produced by the stepwise approach
nonsequential analysis by selecting the best (results not shown). Hence, the final models
subset of CFs that best predicted the outcome, that we presented were robust to the method
we obtained exactly the same set of factors of selection.

Copyright © 2012 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
LWW/IYC IYC200089 February 23, 2012 1:31

Factors Associated With Language Delay 167

DISCUSSION cabulary production, which is not affected by


environmental factors. It does not, however,
The purpose of this study was to exam- apply to receptive language in our sample.
ine potential factors that were related to ex- Children with low receptive language abili-
pressive vocabulary, expressive language, and ties come from environments characterized
receptive language among French-speaking by low parental education and high levels of
children from Quebec diagnosed with LD be- parental stress. Links between education and
tween the ages of 18 months and 36 months. language development have been highlighted
When several factors were considered con- in many studies on expressive language (Hoff
currently, cognitive development was consis- & Tian, 2005; Rowe, 2008). It has indeed been
tently associated with language development. shown that less-educated parents talk less, use
Other child factors such as male gender and a less-varied vocabulary, and provide more
age were also retained in the regression model commands than do more educated parents
explaining expressive vocabulary, whereas (Hoff, 2003). It has also been shown that high
only age was added to cognitive development levels of parental stress predict low expres-
in the model explaining expressive language. sive language in children aged 18–39 months
Two environmental factors were involved in (Horwitz et al., 2003) and low scores in gen-
the development of receptive language, with eral language development in severely ne-
parental level of education and parental stress glected children aged 2–36 months (Sylvestre
accounting for a part of the variance. Taken & Mérette, 2010). However, to the best of our
individually, many factors were not signifi- knowledge, our results are the first to show a
cantly related to LD. The absence of links with specific link between these factors and recep-
positive family history is particularly strik- tive language.
ing, considering the strength of association An important issue raised by our findings
highlighted in recent epidemiological studies is the question as to why factors are differ-
(Reilly et al., 2007; Zubrick et al., 2007). ent across modalities, that is, why parental
Links that were expected between cogni- education and parental stress are related to
tive development and each of the linguistic comprehension and not to production. One
dimensions investigated may at least in part be possible explanation is that, in order for re-
explained by the fact that the mental scale of ceptive language to develop, the child needs
the BSITD-II (Bayley, 1993) comprises many more contextual cues than for expressive lan-
items that test verbal abilities. Nevertheless, guage. When children are confronted with a
these links are not surprising, considering the less-familiar word or statement, for example,
well-known mutual and reciprocal influences they rely on various strategies to compensate
of these two spheres of development. Our re- for their lack of understanding such as cues
sults converge with the point of view that lan- presented within the communication context
guage and cognitive development rest on a (Clark, 2004). To understand a message, chil-
unified developing system using a common dren have to go beyond the meaning of words
set of domain-general learning mechanisms and sentences and grasp speech in real time,
and representational resources (Bates et al., integrate social context language, decode the
2005; Marchman & Thal, 2005). intentions of the persons speaking, and form
With regard to the other factors measured a representation of their communication in-
in this study, the lack of association between tentions (Diesendruck, Markson, Akhtar, &
language development and several factors is Reudor, 2004). The manner in which the con-
not consistent with results of previous reports text supports language comprehension de-
from community samples (Reilly et al., 2006; velopment is compounded by an analysis of
Reilly et al., 2007; Zubrick et al., 2007). These nonlinguistic indications, such as adult ges-
studies suggest a strong biological trajectory tures and glances directed toward a named ob-
for expressive language development and vo- ject (Paul, 2000). It is plausible that elevated

Copyright © 2012 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
LWW/IYC IYC200089 February 23, 2012 1:31

168 INFANTS & YOUNG CHILDREN/APRIL–JUNE 2012

parental stress may affect the child’s availabil- Turning to the caveats of the study, it is
ity and regularity of routines, thereby reduc- clear that the cross-sectional design prevents
ing the quality and quantity of opportunities any inference of the direction of the causal
to put the language and all the stimulation relationships among the phenomena studied.
provided into context (Most, Fidler, Laforce- Indeed, given that the environmental factors
Booth, & Kelly, 2006). identified deal with the child–parent relation-
One must keep in mind that the results of ship, the causality links may be bidirectional.
the present study rely on cross-sectional ob- Moreover, despite the relatively large size of
servations. Given the continuous and longitu- our sample when compared to the corpus of
dinal nature of language development, the ab- studies of LD children younger than 3 years,
sence of a relationship at a given time may not it is also conceivable that it lacked statistical
preclude such a relationship at other points in power to detect links between certain fac-
time. It is plausible, for instance, that children tors and language development. This being
younger than 3 years do not yet differ suffi- said, there are also numerous strengths worth
ciently, in terms of language, to reveal ma- mentioning in this study. The size and qual-
jor influences but that, as they grow and as ity of the sample must be emphasized. Few
their language becomes more complex, the research studies of children younger than 3
links with certain factors may become appar- years with LD have included so many partic-
ent. As language becomes more complex, it ipants, all with a confirmed diagnosis of LD
is also possible that children will be faced by an SLP. Our sample probably consists of
with different types of language difficulties children whose level of development is on a
(Justice, Bowles, Pence, Turnbull, & Skibbe, continuum; that is, some children’s language
2009). Furthermore, the role of different fac- abilities are close to normal limits and others
tors may be significant at certain stages of have more severe problems, which are likely
development, whereas if a factor appeared to lead to a diagnosis of specific-language im-
earlier or later in time, it might have no ef- pairment at a later time. This renders the re-
fect. The timing of the occurrence of the fac- sults directly applicable to the clinical setting.
tor would then be crucial for development. Also, the quantity and variety of factors mea-
In addition, once the factor has occurred, its sured allow simultaneous assessment of all
impact on language may be direct or remain variables known to date to influence the lan-
latent for some time and have a delayed con- guage development of young children. More-
sequence on language. Another possibility is over, it is the first study on children younger
that the duration of exposure to certain fac- than 3 years that has taken an interest in el-
tors may be responsible for their effects. For ements of language development that go be-
example, a child may need to be exposed to yond expressive vocabulary, that is, expres-
major parental depression for some time for sive language in a broad sense as well as
a related effect to become apparent. A dose– receptive language. Finally, the results have
response effect can also be considered in that made it possible to demonstrate that different
the impact of a given factor may become ap- factors are linked to these different linguistic
parent only when it manifests itself with a cer- dimensions.
tain magnitude (e.g., economic deprivation).
Finally, there could be a cumulative effect of CONCLUSION
factors; that is, being exposed to many fac-
tors simultaneously may have more impact This study has given rise to some sugges-
than the nature of the individual factors, as tions for clinical practice and future research
proposed in the model of cumulative risk fac- directions. On the clinical front, we note
tors (Rutter, Pickles, Murray, & Eaves, 2001; that the risk for LD is not equivalent for all
Sameroff, Seifer, Zax, & Barocas, 1987). children because some environmental factors

Copyright © 2012 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
LWW/IYC IYC200089 February 23, 2012 1:31

Factors Associated With Language Delay 169

may have a negative impact on language devel- when the children are still very young. It has
opment. But, most of the early prevention and been shown that early intervention is more
intervention programs target the support and efficient at a younger age (Anderson et al.,
improvement of the quality of parent–child 2003) and that there exist critical develop-
relationships and the quantity and quality ment periods that must not be overlooked
of language stimulation offered by parents to avoid jeopardizing child development even
to the child (Baxendale & Haskett, 2003; further (Thompson, 2001). Our data reinforce
Warr-Leeper, 2001). Despite the fact that the need to act promptly before the delays
these dimensions constitute determining fac- worsen.
tors in child language development, the re- There is still a lot of research work to be
sults of the present study suggest that some done. Environmental factors having an im-
peripheral factors may influence these vari- pact on language development still remain
ables, namely education level and parental somewhat of a mystery. It is essential to con-
stress. duct other cross-sectional studies with sam-
Moreover, the fact that factors associated ples similar to ours in order to validate these
with expressive vocabulary, with a measure first results. Longitudinal studies will also al-
of global expressive language, and with a low researchers to follow the evolution of
measure of receptive language were different child language development and to study the
points to the use of different clinical strate- ways in which the evolution takes place in
gies. This call for the adoption of selective relation to the child’s characteristics and en-
actions, the nature, and intensity should vary vironment. The fact that parental education
according to the analysis of the situation of and parental stress contribute to a negative
the child (Epps & Jacob, 2000). result in terms of comprehension and not in
Finally, given the impact of an LD on sub- expression is of particular relevance for future
sequent child development, it is critical to act research.

REFERENCES

Abidin, R. (1990). Parenting stress index short form— Bishop, D. V. M., Price, T. S., Dale, P. S., & Plomin, R.
test manual. Charlottesville, VA: Pediatric Psychol- (2003). Outcomes of early language delay: II. Etiology
ogy Press. of transient and persistent language difficulties. Jour-
Anderson, L. M., Shinn, C., Fullilove, M. T., Scrimshaw, nal of Speech, Language, and Hearing Research, 46,
S. C., Fielding, J. E., Normand, J., . . . Carande-Kulis, 561–575.
S. (2003). The effectiveness of early childhood devel- Carson, D. K., Klee, T., Perry, C. K., Donaghy, T.,
opment programs. A systematic review. American & Muskina, G. (1997). Measures of language pro-
Journal of Preventive Medicine, 24(3)(Suppl.), ficiency as predictors of behavioural difficulties,
32–46. social and cognitive development in 2-year-old
Bates, E., Tomasello, M., & Slobin, D. (2005). Beyond children. Perceptual and Motor Skills, 84(3, pt 1),
nature-nurture: Essays in honor of Elizabeth Bates. 923–930.
Mahwah, NJ: Erlbaum. Clark, E. V. (2004). How language acquisition builds
Baxendale, J., & Hesketh, A. (2003). Comparison of on cognitive development. TRENDS in Cognitive
the effectiveness of the Hanen Parent Programme Sciences, 8(10), 472–477.
and traditional clinic therapy. International Journal Dale, P., Price, T. S., Bishop, D., & Plomin, R. (2003).
of Language & Communication Disorders, 38(4), Outcomes of early language delay: I. Predicting
397–415. persistent and transient language difficulties at 3 and
Bayley, N. (1993). Bayley Scales of Infant and Toddler 4 years. Journal of Speech, Language, and Hearing
Development (2nd ed.). San Antonio, TX: Harcourt Research, 46(3), 544–560.
Assessment. Desmarais, C., Sylvestre, A., Meyer, F., Bairati, I., &
Bigras, M., LaFrenière, P. J., & Abidin, R. R. (Eds.) Rouleau, N. (2008). Systematic review of the liter-
(1995). Indice de Stress Parental: Complément fran- ature on characteristics of late-talking toddlers. In-
cophone à l’édition américaine. Toronto, Ontario, ternational Journal of Language and Communica-
Canada: Multi-Health System. tion Disorders, 43, 1–30.

Copyright © 2012 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
LWW/IYC IYC200089 February 23, 2012 1:31

170 INFANTS & YOUNG CHILDREN/APRIL–JUNE 2012

Desmarais, C., Sylvestre, A., Meyer, F., Bairati, I., & Justice, L. M., Bowles, R. P., Pence Turnbull, K. L., &
Rouleau, N. (2010). Three profiles of language abil- Skibbe, L. E. (2009). School readiness among chil-
ities in toddlers with an expressive vocabulary de- dren with varying histories of language difficulties.
lay: Variations on a theme. Journal of Speech, Developmental Psychology, 45(2), 460–476.
Language, and Hearing Research, 53(3), 699– Justice, L. M., Invernizzi, M. A., & Meier, J. D. (2002). De-
709. signing and implementing an early literacy screen-
Diesendruck, G., Markson, L., Akhtar, N., & Reudor, A. ing protocol: Suggestions for the speech–language
(2004). Two-year-olds’ sensitivity to speakers’ intent: pathologist. Language, Speech, and Hearing Ser-
An alternative account of Samuelson and Smith. De- vices in Schools, 33, 84–101.
velopmental Science, 7(1), 33–41. Klee, T., Carson, D. K., Gavin, W. J., Hall, L., Kent, A., &
Elman, J. L., Bates, E. A., Johnson, M. H., Karmiloff-Smith, Reece, S. (1998). Concurrent and predictive validity
A., Parisi, D., & Plunkett, K. (1996). Rethinking in- of an early language screening program. Journal of
nateness. Cambridge, England: MIT Press. Speech, Language, and Hearing Research, 41(3),
Epps, S., & Jackson, B. J. (2000). Empowered families, 627–641.
successful children: Early intervention programs Klee, T., Pearce, K., & Carson, D. K. (2000). Improving
that work. Washington, DC: American Psychologi- the positive predictive value of sreening for develop-
cal Association. mental language disorder. Journal of Speech,
Fenson, L., Dale, P., Reznick, S., Thal, D., Bates, E., & Har- Language, and Hearing Research, 43,
tung, J. (1993). MacArthur communicative develop- 821–833.
ment inventories. San Diego, CA: Singular Publishing Kloth, S., Janssen, P., Kraaimaat, F., & Brutten, G. J.
Group. (1998). Communicative styles of mothers interacting
Fournier, L., Lesage, A., Phil, M., Toupin, J., & Cyr, with their preschool-age children: A factor analytic
M. (1997). Telephone surveys as an alternative study. Journal of Child Language, 25, 149–168.
for estimating prevalence of mental disorders and Kovess, V., & Fournier, L. (1990). The DISSA: an abridged
service utilization: A Montreal catchment area self-administered version of the DIS. Approach by
study. Canadian Journal of Psychiatry, 42(7), episode. Society of Psychiatry and Psychiatric Epi-
737–743. demiology, 25(4), 179–186.
Garbarino, J., & Ganzel, B. (2000). The human ecol- Law, J. (1992). The early identification of language im-
ogy of early risk. In: J. P. Shonkoff & S. J. Meisels pairment in children. London, England: Chapman &
(Eds.), Handbook of early childhood intervention, Hall.
(2nd ed. pp. 76–93). New York, NY: Cambridge Uni- Marchman, V., & Thal, D. (2005). Words and gram-
versity Press. mar. In: M. Tomasello & D. Slobin (Eds.), Beyond
Hadley, P. A., & Holt, J. K. (2006). Individual differences nature-nurture: Essays in honor of Elizabeth Bates
in the onset of tense marking: A growth-curve analy- (pp. 141–164). Mahwah, NJ: Erlbaum.
sis. Journal of Speech, Language, and Hearing Re- Most, D. E., Fidler, D. J., Laforce-Booth, C., & Kelly, J.
search, 49, 984–1000. (2006). Stress trajectories in mothers of young chil-
Hampson, J., & Nelson, K. (1993). The relation of mater- dren with Down syndrome. Journal of Intellectual
nal language to variation in rate and style of language Disability Research, 50(pt 7), 501–514.
acquisition. Journal of Child Language, 20, 313– Nelson, H. D., Nygren, P., Walker, M., & Panoscha, R.
342. (2006). Screening for speech and language delay in
Hirsch-Pasek, K., Golinkoff, R. M., Hennon, E. A., & preschool children: Systematic evidence review for
Maguire, M. J. (2004). Hybrid theories at the fron- the US Preventive Services Task Force. Pediatrics,
tier of developmental psychology: The emergentist 117, e298–e319.
coalition model of word learning as a case in point. Paul, R. (2000). “Putting things in context”: Literal and
In: D. G. Hall & S. R. Waxman (Eds.), Weaving a discourse approaches to comprehension assessment.
lexicon. Cambridge, MA: MIT Press. Seminars in Speech and Language, 21(3), 247–255.
Hoff, E. (2003). The specificity of environmental influ- Prathanee, B., Purdy, S. C., Thinkamrop, B.,
ence: Socioeconomic status affects early vocabulary Chaimay, B., Ruangdaraganon, N., & Mo-suwan,
development via maternal speech. Child Develop- L., . . . Phuphaibul, R. (2009). Early language delay
ment, 74, 1368–1378. and predictive factors in children aged 2 years.
Hoff, E., & Tian, C. (2005). Socioeconomic status and Journal of medical Association Thai, 92(7),
cultural influences on language. Journal of Commu- 930–938.
nication Disorders, 38, 271–278. Prathanee, B., Thinkhamrop, B., & Dechongkit, S. (2007).
Horwitz, S. M., Irwin, J. R., Briggs-Gowan, M. J., Bosson Factors associated with specific language impairment
Heenan, J. M., Mendoza, J., & Carter, A. S. (2003). and later language development during early life: A
Language delay in a community cohort of young chil- literature review. Clinical Pediatrics, 46(1), 22–29.
dren. Journal of American Academy of Child and Reilly, S., Eadie, P., Bavin, E. L., Wake, M., Prior, M., &
Adolescent Psychiatry, 42(8), 932–940. Williams, J., . . . Ukoumunne, O. C. (2006). Growth of

Copyright © 2012 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
LWW/IYC IYC200089 February 23, 2012 1:31

Factors Associated With Language Delay 171

infant communication between 8 and 12 months: A Sylvestre, A., & Mérette, C. (2010). Language develop-
population study. Journal of Paediatrics and Child ment of children of less than three years old referred
Health, 42(12), 764–770. for severe parental neglect: Influence of personal and
Reilly, S., Wake, M., Bavin, E. L., Prior, M., Williams, J., environmental factors. Child Abuse & Neglect, 34,
& Bretherton Ukoumunne, O. C. (2007). Predicting 414–428.
language at 2 years of age: A prospective community Sylvestre, A., & St-Cyr Tribble, D. (2001). Traduction au-
study. Pediatrics, 120, e1441-e1449. torisée par l’auteur de l’échelle de développement du
Rescorla, L. (1989). The language development survey: langage. Rossetti-Infant Toddler Language Scales.
A screening tool for delayed language in toddlers. Document inédit. Sherbrooke, Québec, Canada:
Journal of Speech and Hearing Disorders, 54, 587– Université de Sherbrooke.
599. Sylvestre, A., St-Cyr Tribble, D., Payette, H., & Cronk, C.
Rescorla, L., & Merrin, L. (1998). Communicative intent in (1998). Questionnaire de Stimulation de la Com-
late-talking toddlers. Applied Psycholinguistics, 19, munication. Document inédit. Sherbrooke, Québec,
393–414. Canada: Université de Sherbrooke.
Robins, L. N., Wing, J., Wittchen, H. U., Helzer, J. E., Ba- Thompson, B. (1995). Stepwise regression and stepwise
bor, T. F., Burke, J., . . . Regier, D. A. (1988). The discriminant analysis need not apply here: A guide-
Composite International Diagnostic Interview Sim- lines editorial. Educational and Psychological Mea-
plified. An epidemiologic instrument suitable for use surement, 55, 525–534.
in conjunction with different diagnostic systems and Thompson, R. A. (2001). Development in the first years
in different cultures. Archives of General Psychiatry, of life. Future Child, 11(1), 20–33.
45(12), 1069–1077. Trudeau, N., Frank, I., & Poulin-Dubois, D. (1999). Une
Rossetti, L. (1990). Infant Toddler Language Scales. East adaptation en français québécois du MacArthur
Moline, IL: Linguisystems. Communicative Development Inventory. Re-
Rowe, M. L. (2008). Child-directed speech: Relation to vue d’Orthophonie et d’Audiologie, 23(2),
socioeconomic status, knowledge of child develop- 61–73.
ment and child vocabulary skill. Journal of Child Vigil, D. C., Hodges, J., & Klee, T. (2005). Quantity and
Language, 35, 185–205. quality of parental language input to late-talking tod-
Rutter, M., Pickles, A., Murray, R., & Eaves, L. (2001). dlers during play. Child Language Teaching and
Testing hypotheses on specific environmental causal Therapy, 21(2), 107–122.
effects on behavior. Psychological Bulletin, 127(3), Warr-Leeper, G. A. (2001). A review of early in-
291–324. tervention programs and effectiveness for envi-
Sameroff, A. J., Seifer, R., Zax, M., & Barocas, R. (1987). ronmentally disadvantaged children. Journal of
Early indicators of developmental risk: Rochester Speech-Language-Pathology and Audiology, 25(2),
Longitudinal Study. Schizophrenia Bulletin, 13(3), 89–102.
383–394. Weschler, D. (Ed.). (1989). Weschler preschool and pri-
Shonkoff, J. P., & Marshall, P. C. (2000). The biology mary scale of intelligence—revised. San Antonio,
of developmental vulnerability. In: J. P. Shonkoff & TX: The Psychological Corporation.
S. J. Meisels (Eds.), Handbook of early childhood Zubrick, S. R., Taylor, C. L., Rice, M. L., & Slegers, D.
intervention, (2nd ed. pp. 35–53). New York, NY: W. (2007). Late language emergence at 24 months:
Cambridge University Press. An epidemiological study of prevalence, predictors,
Statistical Package for the Social Sciences. Retrieved from and covariates. Journal of Speech, Language, and
http//:www.spss.com Hearing Research, 50, 1562–1569.

Copyright © 2012 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.

Potrebbero piacerti anche