Examining the Associations Between Parent Concerns and School-Age Children's Language and Reading Abilities: A Comparison of Samples Recruited for In-School Versus Online Participation

Abstract

Purpose:

The purpose of this study was to examine the relationship between parent concerns about children's oral language, reading, and related skills and their children's performance on standardized assessments of language and reading, with a particular focus on whether those relationships differed between children recruited for in-school versus online participation.

Method:

This study used data from a larger, longitudinal project focused on children with and without developmental language disorder (DLD) and/or dyslexia. The “in-school” sample (n = 133) completed assessments in-person before school closures, and the “online” sample (n = 84) recruited via advertisements completed assessments online. Parents completed a checklist of concerns. All children completed norm-referenced assessments of language and reading.

Results:

The two recruitment strategies yielded samples that differed in racial diversity (higher in the in-school sample), caregiver education levels (higher in the online sample), and word reading test scores (higher in the online sample). Parents in both samples reported higher levels of concerns about literacy skills than oral language skills, and the correlation between parent concerns about literacy and children's word reading test scores was stronger than the correlation between parent concerns about oral language and children's language test scores.

Conclusions:

Researchers and clinicians should be aware of how recruitment strategies and assessment modalities (e.g., in-person vs. tele-assessment) may impact participation in studies and clinical service. A reliance on parent concerns about oral language to prompt a language evaluation may contribute to low rates of identification of children who meet criteria for DLD. Future research can consider parent concerns about literacy, attention, and executive functions as indicators of a need for language evaluation, especially considering the high comorbidity between language and other developmental disorders.

The sudden shift to remote work in response to COVID-19 restrictions affected clinicians and researchers alike. Clinicians were forced to quickly learn or develop new ways to administer assessments and deliver intervention from a distance, usually over the Internet (Sylvan et al., 2020). For many researchers, moving a research project online involves more than just a change to the medium in which we interact with children. For researchers who typically recruit children from schools for in-person assessments during the school day, it also changes the way that we recruit participants. Furthermore, online data collection requires a different level of parent involvement for scheduling and setting up the environment and technology to allow students to participate in research. Considering these differences, it is reasonable to think that the subset of children that we recruit for participation in research online may be different from those we recruit in schools (Sylvan et al., 2020). This is one aspect of a broader question about the extent to which samples recruited by different sampling methods (e.g., community vs. convenience sampling) represent the broader population of interest, and whether we can generalize findings to the broader population (Bornstein et al., 2013). The current study examines the relationship between parent concerns about children's oral language, reading, and related skills and their children's performance on standardized assessments of language and reading, with a particular focus on whether those relationships differed between children we recruited for in-school (community sampling) versus online participation (convenience sampling).

The study took place in the context of a larger project focused on children with impairments in spoken language (e.g., developmental language disorder [DLD]) and/or word reading (e.g., dyslexia). Consistent with the Simple View of Reading (Gough & Tunmer, 1986), deficits in either of these areas convey risk to children's reading comprehension, but reading comprehension problems can manifest at different points of academic development. Word reading abilities are strongly predictive of reading comprehension in the early grades, whereas oral language skills (including vocabulary, semantics, syntax, and inference) are more predictive of reading comprehension in later school grades where the linguistic demands are higher (Foorman et al., 2018). Thus, the word reading difficulties involved in dyslexia are likely to have a functional impact on reading comprehension beginning in the early school years (Chapman & Tunmer, 2003; McArthur et al., 2000). In contrast, the impact of language disorders on reading comprehension may not be observable until somewhat later, especially for those children who develop adequate word reading abilities (Catts et al., 2012; Fong & Ho, 2019; Justice et al., 2013; Lipka et al., 2006; Nation et al., 2010). Thus, the impacts of DLD and dyslexia are expected to be different through the school years (Snowling et al., 2019).

Past research suggests that there is less public awareness of oral language difficulties compared with literacy difficulties, including among parents (Adlof, 2020; Bishop, 2010; McGregor, 2020). One reflection of low public awareness of language difficulties is proportionately low rates of research about DLD. McGregor (2020) reported the number of peer-reviewed journal articles published on DLD versus dyslexia for the years 2000–2019. Although these two disorders have similar prevalence rates, there have been more than twice the number of publications on dyslexia than DLD. In terms of parents specifically, Adlof et al. (2017) found that parents of children with dyslexia, with and without DLD, reported concerns and/or prior history of services at significantly higher rates than children with DLD only (56.2% and 60% vs. 29%, respectively). The authors interpreted these results as an indication that parents of children across the three impairment groups may be less aware of children's oral language difficulties than their reading difficulties (see also Hendricks et al., 2019).

Low levels of public awareness about oral language development and disorders are important because access to speech and language services typically relies on referrals, initiated by concerns from parents, teachers, or other professionals about a child's language development. Language weaknesses may present differently in social versus academic settings, which have differing linguistic demands, making language difficulties difficult to detect without direct measurement. For example, the impact of language disorders may be less evident in the conversational language often used at home compared with the more decontextualized oral and written language that is used in schools (Snow & Uccelli, 2009; Uccelli et al., 2019), making it more difficult for parents to notice language difficulties. Although language disorders impact school performance, many teachers cannot identify features of language impairment in school-age children (Antoniazzi et al., 2009; Christopulos & Kean, 2020; McLeod & Harrison, 2009; Williams, 2006). In this case, teachers may not initiate conversations with parents about possible language concerns. In contrast, reading difficulties may be easier to detect than oral language difficulties because reading is directly taught and monitored in schools.

In the clinical context, low rates of identification of language impairments raise concerns about the number of children who are not identified and not receiving academic supports (Norbury et al., 2016; Tomblin et al., 1997). In research contexts, recruitment methods that rely on referrals or parent concerns may produce different samples than those that use community-based methods. This is because identification rates for DLD vary with many factors, including gender, race, socioeconomic status, and presence of factors including speech sound disorders, attention, and executive function deficits (McGregor, 2020; Morgan et al., 2016; Redmond, 2016; Wittke & Spaulding, 2018; Zhang & Tomblin, 2000). Furthermore, rates of comorbidity of language and reading impairment are generally higher in studies that use clinical referrals and convenience sampling than community-based sampling approaches (Catts et al., 2005; Dewey, 2021; McArthur et al., 2000).

In the case of the current study, the COVID-19 pandemic led to a change in recruitment and data collection procedures for our larger project. Prior to COVID-19 restrictions, recruitment for our larger project primarily involved a community-based sampling approach in which all second grade students in participating schools were initially invited to participate, and all data collection was conducted in person during the school day. In response to school closures, we shifted to collecting data online over videoconferencing software. We also shifted to a convenience sampling approach and recruited participants from advertisements that targeted children with suspected reading or language difficulties as well as those with typical development. The shift in procedures impacted the level of parent involvement required for children to participate in the study. For in-person, in-school data collection, minimal parent involvement was required other than the return of the signed consent form and parent questionnaires. Researchers worked with school administrators and teachers to arrange student testing schedules, and the data collectors worked with students individually on site. For online data collection, parents were the primary contact for all communication, including scheduling multiple appointments for their child and setting up equipment for data collection on the participant's end (e.g., computer or tablet with a camera in a quiet environment). This presented a unique opportunity to study differences between samples of children recruited via community sampling (in schools) versus convenience sampling (advertisements in a variety of mediums). Such differences are of importance not only to this study but also to the interpretation of other studies about DLD that use convenience sampling. We were especially interested in levels of parental concern, and the extent to which those concerns match measured language and reading abilities. This is important because when children's participation in research or clinical intervention programs requires high levels of parental initiative, the levels of parental concern for those children who participate may be different from the population as a whole.

This study addressed three research questions: First, considering reading, language, and related skills (e.g., articulation, attention, and executive functioning), what are the most commonly reported areas of concern by parents in the school-based versus online samples? Based on results of past studies (e.g., Hendricks et al., 2019; McGregor, 2020), we hypothesized that parents would be more likely to report concerns about children's literacy skills than their oral language abilities. Considering the higher level of parent involvement required for online versus in-school participation, and the fact that we advertised for children with language or reading difficulties, we hypothesized that parents in the online sample would report more concerns overall than parents in the school sample. Second, what are the associations between parent concerns and children's language and word reading abilities as measured by norm-referenced assessments? We hypothesized that parent concerns and children's performance on standardized tests would be more strongly correlated for word reading than oral language because word reading difficulties may be more visible than oral language difficulties. Third, do the associations between parent concerns and children's language and word reading scores differ between the two samples? Following from our second hypothesis in which we expected higher levels of parent concern in the online sample compared with the in-person sample, we also hypothesized that parent concerns and children's language and word reading scores would be more strongly correlated in the online sample than in the school sample.

Method

This study used data from a larger, longitudinal project focused on children with DLD, dyslexia, both disorders, or typical development. Data were collected at the University of South Carolina and the University of Pittsburgh, and all study procedures were approved by the institutional review boards at each site.

Participants

Participants included a total of 217 second grade students recruited through two sampling methods. The “in-school” sample (n = 133) included children recruited from 12 classrooms across three schools in one school district in South Carolina in the 2018–2019 school year (n = 76), and 10 classrooms in two schools in the 2019–2020 school year before the COVID-19 school closures (n = 57). The study information provided to parents indicated that the study would involve children with and without language and/or reading impairments. One of the schools participated both years. For the in-school sample, we distributed study information packets with paper consent forms to parents/caregivers (“parents” will be used in all future instances) of all second graders. Parents were instructed to complete and return the consent form to their classroom teacher to be returned to the research team. Those who spoke English as their primary language were invited to participate in a larger, ongoing study in which all data collection was conducted in person during school hours. Classrooms were given monetary incentives to acknowledge the participation of their students.

Participants in the “online” sample (n = 84) were recruited via advertisements posted on local and national research participant registries; on American Speech-Language-Hearing Association Special Interest Groups message boards; on websites and e-mail lists for the South Carolina Speech Language Hearing Association and Georgia Speech-Language-Hearing Association; on e-mail lists for local public libraries and tutoring centers for children with reading difficulties; and on social media groups for parents, educators, and speech-language pathologists (SLPs) that were organized around topics such as dyslexia, the science of reading, evidence-based practice in speech-language pathology, and local and regional school district issues. Advertisements targeted children with reading or language difficulties as well as typically developing children who spoke English as their primary language, and all advertisements were sent with requests to reshare. The online sample had the option to e-mail a signed consent form to the project coordinator or complete their signed consent electronically using research electronic data capture (REDCap) tools hosted at the University of South Carolina and University of Pittsburgh (Harris et al., 2009, 2019). The online sample includes participants from 17 states in the United States who each received a monetary incentive for participating in the study.

For all children included in this study, parents reported no history of autism, uncorrected vision, or motor disorder (including apraxia of speech). Parents of a few children who were included in the study reported other diagnoses such as attention (attention-deficit/hyperactivity disorder [ADHD]/attention deficit disorder [ADD]; n = 11 or 8.3% in-school; n = 9 or 10.7% online), dyscalculia (n = 1 or 0.8% in-school), slow processing disorder (n = 1 or 0.8% in-school), and anxiety disorder (n = 1 or 1.2% online). Standardized assessments of language and reading ability were administered individually in person or on Zoom, depending on the sample and were generally completed first in a series of participant sessions.

Measures

Parent Questionnaire

Parents completed an intake questionnaire requesting demographic information as well as information about their child's medical and educational history, school lunch status (i.e., whether the child received free or reduced price school lunch), family history of language or reading disabilities, and history of prior support services, in addition to a checklist of concerns related to language, literacy, and related skills. The question about history of prior support services differed in the 2018–2019 (in-school sample) and 2019–2020 parent questionnaire (in-school and online samples). In 2018–2019, the question was worded as “Has your child ever received speech, language, reading or other special education services? If yes, please describe.” In 2019–2020, the question was presented as “Has your child received any of the following reading/math services? If yes, for how long?” The parent was instructed to identify length of participation (e.g., “Never” to “3+ years”) for each type of reading/math service (e.g., private school, response to intervention, summer camp) Here, we report the percentage of families who reported receiving any services across these question formats.

The parent concerns checklist includes a list of concerns that may indicate a risk for reading or language difficulties. It was adapted from the work of Hendricks et al. (2019) by adding two items related to executive functioning. Parents were asked “Are you concerned about your child's performance in any of the following areas? Please check all that apply.” Twelve items assessed parent concerns related to oral language, speech articulation, literacy, attention, executive functions, and “other” (see Table 3 for specific items). Parents indicated an area of concern by checking each area of concern for their child and wrote any additional concerns under “other.” Parents of in-school participants completed the questionnaire in paper form, and parents of online participants completed the questionnaire using REDCap.

Table 3.

Reports of parental concerns for children recruited through in-school sample and online sample.

Item	In-school (n = 133)	Online (n = 84)
Percentage (%) of parents reporting concerns
Understanding what you tell him/her at home a	6.8	2.4
Understanding teachers at school a	9	1.2
Expressing his or her thoughts when speaking a	9	3.6
Reading individual words b	12.8	11.9
Spelling b	15	20.2
Understanding what he/she reads b	26.3	14.3
Writing sentences or longer texts b	18.8	21.4
Saying words correctly	12	9.5
Paying attention	27.1	25
Remembering instructions or personal belongings	24.1	15.5
Following routines	3.8	7.1
Other concerns	3.0	2.4
Number of reported concerns (max = 12), M (SD)	1.68 (2.22)	1.35 (2.14)
Mean of oral language concerns	.08 (.20)	.02 (.13)
Mean of literacy concerns	.18 (.30)	.17 (.30)

Note. In terms of parent reported concerns, parents/caregivers indicated Yes/No on a checklist.

^a Oral language concerns.

^b Literacy concerns.

Oral Language

The Clinical Evaluation of Language Fundamentals–Fifth Edition (CELF-5; Wiig et al., 2013) was administered as a comprehensive measure of language ability assessing domains including semantics, morphology, and syntax. It is normed for students between 5 and 21 years old. The standardization sample was selected to be representative of the population of the United States, and the test development process included several steps to reduce cultural bias (i.e., expert review and psychometric tests for item bias). CELF-5 norms for children in the 8;0–8;11 (years;months) age range were derived from a sample in which 12.5% of children identified as African American, 4% as Asian, 19.5% as Hispanic, 56.5% White, and 7.5% as “Other.” The Core Language Score (CLS) represents a composite of scores on the Sentence Comprehension, Word Structure, Formulated Sentences, and Recalling Sentences subtests for children up to 8 years old, and Word Classes, Semantic Relationships, Formulated Sentences, and Recalling Sentences for students between 9–21 years old. The CELF-5 technical manual reports strong reliability estimates for the CLS ranging from .95 to .97 for children aged 7:0–9;11. The clinical validity statistics show high sensitivity (1.00) and specificity (.91) for students with language disorders based on a CLS of 1 SD below the mean.

Word Reading

The Word Identification and Word Attack subtests of the Woodcock Reading Mastery Test–Third Edition (WRMT-III; Woodcock, 2011) were administered to assess children's ability to read real words and decode nonwords, respectively. The Basic Skills score is a composite of scores on these two tests, with norms available for individuals aged 4;6 to 79;11. The WRMT-III standardization sample was selected to be representative of the population of the United States, and the test development process included the same checks for cultural bias as the CELF-5. WRMT-III norms for children in the 8;0–8;11 age range were derived from a sample in which 14% of children identified as African American, 4% as Asian, 20% as Hispanic, and 4% as other. The WRMT-III technical manual reports strong estimates for internal consistency for the Basic Skills Cluster at .95 (Form A) and high content and construct validity as demonstrated by relations between the WRMT-III subtest scores, cluster scores, and other reading assessments and large, statistically significant differences between the population sample and a clinical sample of children with word reading difficulties.

Online Administration

To assess participants online, we used teleconferencing software and digital testing materials. We chose Zoom as our teleconferencing software due to its screen sharing capabilities and ease of use without formal training. We purchased access to digital stimulus books for the CELF-5 and WRMT-III through Pearson's Q-Global Platform. Pages from the digital stimulus books (CELF-5 or WRMT-III) were shown to online participants via screen sharing. The task directions and expectations for responses were largely consistent with in-person administration for three of the four CELF- subtests and the WRMT-III.

In-person administration of the CELF-5 Sentence Comprehension involves participants pointing to their answer on a page, which created a challenge for online administration. Pearson provided some guidance for the CELF-5 Sentence Comprehension involving (a) the assessor sharing screen, (b) the participant providing a response by pointing to their screen using their finger, and (c) the parent or the participant holding another camera to view where the participant pointed for scoring. We trialed this method with a few participants at one site (University of Pittsburgh), but challenges were the additional effort required from the parent (e.g., to be present, support technological issues) and the need for access to another device with a video camera. To remove these barriers to participation, we began using the annotation feature in the Zoom software. Both sites trialed two ways of using the annotation feature: (a) participant annotation in which the participant indicated their response by stamping a checkmark in the appropriate box corresponding to their answer for each item and (b) assessor annotation in which the assessor used the annotation feature to add letters to each answer choice box, and the participant stated the letter of their answer verbally. The University of Pittsburgh site had greater success with participant annotation and used this as the primary method for CELF-5 Sentence Comprehension administration; however, assessor annotation was used for participants with a Chromebook as their device because the interface of Zoom made participant annotation less reliable. At the University of South Carolina site, assessor annotation was the primary method used for all participants.

Results

Table 1 lists the descriptive statistics for both samples. The samples were similar in terms of age and gender. Both samples included a majority of White and Non-Hispanic participants. There was a greater proportion of participants who identified as Black or African American in the in-school sample than the online sample, and this difference was statistically significant when we compared the proportion of families who reported their race as “White” or “Non-White,” Χ ²(1) = 4.06, p = .044. A higher number of parents in the in-school sample indicated eligibility for free or reduced-price school lunch (42.9%) compared with the online sample (14.3%), and this difference was significant, Χ ²(1) = 21.76, p < .001. For both samples, “Mother” represented the majority of responses for Caregiver 1. The most common response for Caregiver 2 in both samples was “Father,” but this represented a smaller proportion of the in-school sample than the online sample, 57.9% versus 83.3%; Χ ²(4) = 19.34, p < .001. Although there was some overlap across samples, caregivers in the online sample tended to report higher levels of education than caregivers in the in-school sample. Level of education ranged from 1 = Less than high school to 6 = Master's degree or higher. We used Mann–Whitney U tests to examine group differences. For both Caregivers 1 and 2, level of education was significantly higher in the online sample compared with the in-school sample, with Caregiver 1 (U = 1726.50, p < .001) having a medium to large effect size (r = .41) and Caregiver 2 (U = 1513.00, p < .001) also having a medium to large effect size (r = .40).

Table 1.

Demographics of participants across sites.

Variable	In-school (n = 133)	Online (n = 84)	Test statistic (df)	p
Age in months, M (SD)	97.72 (4.86)	95.08 (5.72)
Gender (%)			Χ 2(2) = 3.36	.19
Female	50.4	46.4
Male	49.6	51.2
Other	0	2.4
Race (%) a			Χ 2(1) = 4.06	.04
American Indian/Alaska Native	0.8	0
Asian	0.8	1.2
Black or African American	20.3	8.3
Native Hawaiian or Other Pacific Islander	0.8	0
White	69.9	82.1
Other	4.5	4.8
More than one race	3.0	3.6
Ethnicity (%)			Χ 2(1) = .004	.95
Hispanic or Latino	4.5	4.8
Not Hispanic or Latino	86.5	95.2
Unknown/not reported	9.1	0
School lunch
Free or reduced-price school lunch	42.9	14.3	Χ 2(1) = 21.76	< .001
Paid lunch	50.4	83.3
Caregiver 1 is a			Χ 2(3) = 1.33	.72
Mother	88.7	91.7
Father	5.3	4.8
Other	3.8	2.4
Unknown/not reported	2.3	1.2
Caregiver 1 education (%)			U = 1726.50	< .001
Less than high school	3	0
High school diploma/general educational development	18.8	3.6
Some college	30.1	4.8
Associate's degree/technical certification	17.3	6
Bachelor's degree	15.8	20.2
Master's degree or higher	11.3	64.3
Unknown/not reported	3.8	1.2
Caregiver 2 is a			Χ 2(4) = 19.34	< .001
Mother	3.8	6
Father	57.9	83.3
Other	22.6	3.6
Unknown/not reported	15.8	7.1
Caregiver 2 education (%)			U = 1513.00	< .001
Less than high school	6.8	1.2
High school diploma/general educational development	27.8	8.3
Some college	17.3	2.4
Associate's degree/technical certification	13.5	7.1
Bachelor's degree	12.8	29.8
Master's degree or higher	5.3	44
Unknown/not reported	16.5	7.1
Percentage of parents reporting
History of services (2018–2019)b	18.4	N/A
History of reading/math services (2019–2020)c	33.3	29.8	Χ 2(1) = .161	.69
Family history of language or reading disabilities	17.3	16.7	Χ 2(1) = .104	.75

Note. N/A = not applicable.

^a The in-school sample was more racially diverse than the online sample when comparing proportion of students who reported their race as “White” or “Non-White” (combined: American Indian/Alaska Native, Asian, Native Hawaiian or other Pacific Islander, other, and more than one race).

^b The question about history of prior support services differed in the 2018–2019 (in-school sample) and 2019–2020 parent questionnaire (in-school and online samples). In 2018–2019, the question was worded as “Has your child ever received speech, language, reading, or other special education services? If yes, please describe.”

^c In 2019–2020, the question was presented as “Has your child received any of the following reading/math services? If yes, for how long?” The parent was instructed to identify length of participation (e.g., “Never” to “3+ years”) for each type of reading/math service (e.g., private school, response to intervention, summer camp) We report the percentage of families who reported receiving any services across these question formats. No significant differences between samples in history of reading/math services were found.

In terms of history of services, we report the percentage of families who reported receiving any services across those two years. In 2018–2019, 18.4% of the in-school sample reported that they received speech, language, reading, or other special education services. For 2019–2020, the in-school sample had similar reported levels of history of reading and/or math services (33.3%) compared with the online sample, 29.8%; Χ ²(1) = .161, p = .688. The percentage of participants with a family history of language or reading disabilities were also similar for the in-school (17.3%) and online samples, 16.7%; Χ ²(1) = .104, p = .748.

Table 2 shows the descriptive statistics for norm-referenced measures of word reading and oral language abilities for each sample. Both samples scored near the normative mean on the CELF-5 Core Language and were not significantly different from each other, t(215) = −1.005, p = .32. Considering word reading abilities, the mean for the in-school sample was significantly lower than the online sample, who scored near the normative mean, t(215) = −3.243, p = .001, and the effect size was medium (d = −.45, p < .01). We also calculated the proportion of each sample that could be considered at risk for DLD or dyslexia, using a 1 SD below the mean cutoff on the CELF-5 or the WRMT-III, respectively. As shown in Table 2, these proportions were larger in the in-school sample (19.6% oral language; 26.3% word reading) than the online sample (11.9% oral language; 13.1% word reading).

Table 2.

Descriptive statistics for children's language and literacy abilities.

Variable	In-school	Online	t	p
CELF-5 Core Language Index Score, M	98.92	100.90	−1.005	.32
SD	13.89	14.66
Range	70–146	57–137
Number (%) of children scoring 1 SD below mean	26 (19.6%)	10 (11.9%)
WRMT-III Basic Skills standard score, M	95.47	102.18	−3.243	.001
SD	14.45	15.42
Range	64–135	67–134
Number (%) of children scoring 1 SD below mean	35 (26.3%)	11 (13.1%)

Note. CELF-5 = Clinical Evaluation of Language Fundamentals–Fifth Edition; WRMT-III = Woodcock Reading Mastery Test–Third Edition.

Parent Concerns About Reading, Language, and Related Skills by Sample

Our first research question focused on the commonly reported areas of concern by parents related to reading, language, and related skills in the in-school versus online sample (see Table 3). Overall, the rate of concerns in any category was relatively low across both samples. Slightly fewer than half of parents in the in-school sample (43.6%) and just over half of parents in the online sample (54.8%) reported no concerns in any area. Of the concerns that were reported, “paying attention” was the highest reported concern by parents for both samples (27.1% in-school; 25% online). For the in-school sample, the second highest concern was “Understanding what he/she reads” followed by “Remembering instructions or personal belongings.” For the online sample, the second highest concern was “Writing sentences or longer texts” followed by “Spelling.”

We hypothesized that parents would be more likely to report concerns about literacy than oral language. Consistent with this hypothesis, parents in both samples reported concerns related to receptive and expressive oral language less often (6.8%–9% in-school; 1.2%–3.6% online) than concerns about word reading (12.8% in-school; 11.9% online), spelling (15% in-school; 20.2% online), writing (18.8% in-school; 21.4% online), and reading comprehension (26.3% in-school; 14.3% online). To further examine this hypothesis, we computed an average score for the three oral language concerns including “Understanding what you tell him/her at home,” “Understanding teachers at school,” and “Expressing his or her thoughts when speaking”; and an average score for literacy-related concerns including “Reading individual words,” “Spelling,” “Understanding what he/she reads,” and “Writing sentences or longer texts.” Wilcoxon signed-ranks tests confirmed that average parental concern about literacy was significantly higher than average parental concern about oral language in both the in-school sample (z = −3.89, p < .001) with a small to medium effect size (r = −.24) and the online sample (z = −4.32, p < .001) with a small to medium effect size (r = −.26).

We also hypothesized that parents in the online sample would report more concerns on the checklist overall than parents in the in-school sample. To calculate number of reported parent concerns, we calculated an average out of 12 on the checklist (see “Number of reported concerns” in Table 3). There was no significant group difference detected (U = 4963.00, p = .141) on number of reported concerns. We further examined differences in averaged language versus literacy concerns. For language concerns, the parents of the in-school sample had significantly more concerns than the online sample (U = 4832.50, p = .003) with a small effect size (r = −.14). For literacy concerns, no significant difference was found between groups (U = 5400.50, p = .619). Thus, the hypothesis of higher rates of concerns for the online sample was not supported, whether considering total concerns, language-related concerns, or literacy-related concerns. Overall, parents in both samples reported significantly more concerns related to literacy than oral language.

Relations Between Parent Concerns and Children's Language and Reading Abilities

To address research questions 2 and 3, we examined the associations between parent-reported concerns on the intake questionnaire and children's abilities on norm-referenced measures of language and word reading for each sample (see Table 4). We hypothesized that the correlation between parent concerns about children's literacy skills and their word reading scores would be higher than the correlation between parent concerns about children's oral language abilities and their oral language scores. We also hypothesized that the relationship between parent concerns and children's scores on reading and language tests would be stronger in the online sample than the in-school sample. First, we considered the alignment of parent concerns about oral language abilities and children's scores on oral language tests. The correlation between mean oral language concerns and CELF-5 CLSs was significantly stronger in the in-school sample than the online sample (in-school r = −.28, p < .01; online r = .04, p > .05; z = −2.315, p = .01). Note that the negative correlation indicates that higher levels of parent concerns about oral language were associated with lower scores on the CELF-5 Core Language. The lack of a significant correlation in the online sample is due to very low rates of reported oral language concerns in the online sample.

Table 4.

Correlations between parent concerns and children's language and reading abilities.

	In-school (n = 133)					Online (n = 84)
Variable	1	2	3	4	5	1	2	3	4	5
1. Total parent concerns	—					—
2. Mean oral language concerns	.80***	—				.64***	—
3. Mean literacy concerns	.82***	.49***	—			.89***	.45***	—
4. CELF-5 Core Language Score	−.35***	−.28**	−.38***	—		−.28**	.04	−.37**	—
5. WRMT-III Basic Skills	−.33***	−.19*	−.41***	.59***	—	−.36**	.002	−.44***	.61***	—
Parent concerns
Understanding what you tell him/her at home	.55***	.75***	.28**	−.20*	−.18*	.60***	.94***	.44***	.09	.02
Understanding teachers at school	.65***	.80***	.42***	−.27**	−.19*	.50***	.80***	.31**	−.04	−.04
Expressing his or her thoughts when speaking	.52***	.62***	.35***	−.14	−.04	.60***	.92***	.43***	.03	.01
Reading individual words	.59***	.30**	.72***	−.20*	−.31***	.65***	.39***	.75***	−.30**	−.41***
Spelling	.65***	.39***	.82***	−.28**	−.32***	.74***	.28*	.86***	−.31**	−.46***
Understanding what he/she reads	.66***	.41***	.77***	−.37***	−.30***	.70***	.44***	.74***	−.32**	−.26*
Writing sentences or longer texts	.71***	.45***	.88***	−.34***	−.37***	.73***	.34**	.83***	−.26*	−.29**
Saying words correctly	.48***	.39***	.33***	−.15	−.18*	.56***	.45***	.33**	−.02	−.16
Paying attention	.53***	.37***	.19*	−.11	−.05	.69***	.31**	.50***	−.19	−.24*
Remembering instructions or personal belongings	.66***	.62***	.26**	−.18*	−.10	.60***	.25*	.37**	−.17	−.16
Following routines	.35***	.18*	.14	−.03	−.11	.56***	.30**	.35**	−.21	−.24*
Other concerns	.01	−.07	−.11	.19*	.08	−.03	−.03	−.09	.14	.14

Note. CELF-5 = Clinical Evaluation of Language Fundamentals–Fifth Edition; WRMT-III = Woodcock Reading Mastery Test–Third Edition.

^* p < .05.

^** p < .01.

^*** p < .001.

Turning to reading abilities, the mean parent concerns about literacy was significantly and moderately correlated with WRMT-3 Basic Skills scores in both samples (in-school r = −.41, p < .001; online r = −.44, p < .001; z = 0.259, p = .40). Likewise, the correlations between individual checklist items related to literacy and WRMT-3 Basic Skills scores were of similar magnitudes for the two samples. Consistent with our hypothesis, the correlations between literacy concerns and WRMT-3 Basic Skills scores were numerically larger than the correlations between concerns about oral language and CELF-5 CLSs for both samples; however, this difference was only significant for the online sample (z = 3.26, p = .001; in-school z = 1.19, p = .12).

It is also important to consider the significant positive, moderately high correlation between children's language and word reading scores, reflecting shared variance between these skills (r = .59, p < .001 in-school; r = .61, p < .001 online). Given this relationship, parent concerns about literacy may indicate potential weakness in oral language and vice versa. Indeed, mean literacy concerns were significantly correlated with CELF-5 CLSs for both samples, with a similar magnitude (r = −.38, p < .001 and r = −.37, p < .01, z = 0.082, p = .47) that was numerically higher than the correlation between CELF-5 CLSs and mean language concerns for both samples, this difference was only significant for the online sample (online: z = 2.73, p = .003; in-school: z = 0.906, p = .18). On the other hand, mean literacy concerns were more strongly correlated with WRMT-3 Basic Skills scores than mean language concerns for both samples (in-school: z = 1.961, p = .025; online: z = 3.02, p = .001). Overall, these results suggest a greater alignment between parent concerns about literacy and both reading and language scores than between parent concerns about oral language and oral language scores.

Next, we examined correlations between total parent concerns and language and reading scores. Overall, the strength of these associations was similar across both samples. The total number of parent concerns was correlated with CELF-5 CLSs at r = −.35 (p < .001) for the in-school sample and r = −.28 (p < .01) for the online sample. The total number of parent concerns was correlated with WRMT-3 Basic Skills scores at r = −.33 (p < .001) for the in-school sample and r = −.36 (p < .01) for the online sample.

Overall, the results showed few differences between the in-school and online samples in the relation between parent concerns and children's performance on oral language and reading assessments. Therefore, as a complement to our correlational analyses, we combined the two samples to examine the rate of parent concerns in children who would be considered at risk for oral language or reading impairment, as evidenced by scores of at least 1 SD below the mean. Table 5 shows that compared with parents of typically developing children, a greater proportion of parents of children who scored below the cutoff for either oral language or word reading reported concerns on each item of the checklist except “Other.” The total number of concerns was also significantly higher for parents of children with low oral language than parents of typically developing children, t(193) = −4.34, p < .001, d = .80, and for parents of children with low word reading compared with parents of typically developing children, t (203) = −4.07, p < .001, d = .68, p < .001. Parents of children in both at-risk groups were more likely to report literacy concerns than oral language concerns. However, the majority of parents of children scoring below the cutoff for either oral language or word reading reported “no” for each of the areas of concern on the checklist, suggesting that the checklist was relatively insensitive for identifying children at risk for difficulties with oral language or word reading.

Table 5.

Reports of parental concerns for typically developing children and children scoring at least 1 SD below the mean on oral language or word reading across the combined in-school and online samples.

Item	Typically developing (n = 159)	Low oral language (n = 36)	Low word reading (n = 46)
Percentage (%) of parents reporting concerns
Understanding what you tell him/her at home a	3.8	8.3	10.9
Understanding teachers at school a	3.8	13.9	10.9
Expressing his or her thoughts when speaking a	5	16.7	10.9
Reading individual words b	8.2	25	28.3
Spelling b	10.7	41.7	34.8
Understanding what he/she reads b	15.1	44.4	41.3
Writing sentences or longer texts b	13.2	41.7	37
Saying words correctly	9.4	11.1	17.4
Paying attention	23.3	41.7	34.8
Remembering instructions or personal belongings	18.9	30.6	28.3
Following routines	3.8	8.3	8.7
Other concerns	3.8	0	0
Number of reported concerns (max = 12), (M SD)	1.19 (1.83)	2.83 (2.81)	2.63 (2.89)
Mean of oral language concerns	.04 (.15)	.13 (.26)	.11 (.23)
Mean of literacy concerns	.12 (.25)	.38 (.38)	.35 (.39)

Note. In terms of parent reported concerns, parents/caregivers indicated Yes/No on a checklist.

^a Oral language concerns.

^b Literacy concerns.

Discussion

With the rise in the number of studies utilizing tele-assessment to study clinical populations, including young children (e.g., Tambyraja et al., 2021), it is important to consider how recruitment methods influence sample attainment. The nature of tele-assessment and the level of involvement required of families necessitates that only a subset of children who could be typically assessed in schools will be able to participate in online research. Such issues may be particularly important for research focused on “hidden” disorders, such as DLD, which are known to be underidentified, yet primarily rely on referrals for identification in both research and clinical practice. Past studies of DLD that have recruited from clinical caseloads or used convenience sampling have shown higher rates of word reading impairment in those samples than studies that used community-based sampling (Dewey, 2021). With this context in mind, we examined parent concerns about reading, language, and related skills across two samples, an in-school sample (community sample) and an online sample (convenience sample), and investigated the associations between parent concerns and children's performance on norm-referenced measures of language and word reading ability. As we predicted, parents reported higher levels of concerns about literacy skills than oral language skills in both samples, and the correlation between parent concerns about literacy and children's word reading test scores was stronger than the correlation between parent concerns about oral language and children's language test scores. However, we did not observe higher level of concerns in the online sample compared with the in-school sample or find stronger associations between parent concerns and children's reading and language scores in the online sample than the in-school sample.

The in-school sample was recruited with a community sampling approach. All second grade students enrolled in the participating schools were invited to participate, and all who met broad inclusionary criteria (i.e., spoke English as their primary language; no medical history that would be expected to impact reading or language development) were included in the study. The community represented by this sample is drawn from a single school district, whereas our online sample was recruited nationally through targeted advertisements and word of mouth. Although we made efforts to recruit participants with a range of backgrounds and were successful in recruiting participants with a wide range of oral language and word reading abilities, the two samples differed in several ways. First, there was somewhat more racial diversity among participants and more variability in who served as caregivers in the in-school sample than in the online sample. Second, the average participant in the online sample had higher socioeconomic status, as indicated by school lunch status and caregiver educational attainment, than the average participant in the in-school sample. This is not surprising considering that the online sample was recruited primarily through digital advertisements, and that there was a greater level of parent involvement for online participation. We also consider the time period of recruitment, during the COVID-19 pandemic, as another contributing factor to differences in sample demographics. Although all families were required to adjust to changing demands of the pandemic, research suggests that health, economic, and employment impacts varied by race and ethnicity (e.g., Fisher et al., 2020; Gregory et al., 2020; Singu et al., 2020). We further acknowledge that families with lower financial resources may have faced additional barriers to participation during this time. Despite these demographic differences, we, nonetheless, found that the mean language scores in both samples fell within the average range and were not different from each other. In contrast, there was a significant difference between groups for word reading scores. Both group means fell within normal limits, but the in-school sample mean was in the low-average range and significantly lower than the online sample mean that was near the normative mean. Although we cannot draw strong conclusions about these sample differences as they relate to language or reading impairment with the data we have, they highlight the importance of considering how recruitment methods may influence sample attainment.

We first investigated the rates and types of concerns reported by parents in each sample. The rate of concerns in each category was low overall, and the total number of concerns reported by parents did not differ significantly between groups. Approximately half of the parents in each sample reported no concerns in any area, which is consistent with the finding that most children scored within normal limits on both language and word reading assessments. As we predicted, parents in both samples reported literacy concerns significantly more often than language concerns. However, the single most common parent concern in both samples was related to attentional processes. A few children in both samples had a prior diagnosis of ADD; however, even when these children were excluded in a separate analysis, concerns about attention remained the highest reported area of concern in both samples. Similar results were reported by Hendricks et al. (2019) and may be related, in part, to a greater public awareness of ADD/ADHD as compared with DLD (Bishop, 2010).

Next, we considered the alignment between parent concerns about literacy and oral language and children's performance in these areas. Across both samples, we found that parent concerns about literacy were more strongly correlated with children's performance in word reading and oral language than parent concerns about oral language were. Indeed, for the online sample, there was a near-zero correlation between concerns about oral language and children's oral language test scores. This was due to the low rate of concerns about oral language in the online sample, although the range of scores included several children who scored in the range of language impairment. Overall, these results extend those of past studies that found lower levels of parent concerns about oral language than reading in children with DLD (Hendricks et al., 2019) and lower rates of parent concerns in general for children with oral language difficulties than those with word reading difficulties (Adlof et al., 2017). Note that we did not subdivide our samples into subgroups with single versus combined word reading and/or language difficulties given our relatively modest sample size, so the values reported in Table 5 are relatively similar for the two at-risk groups. Taken together, these findings support ongoing efforts to raise public awareness about DLD, which has important impacts on both social and academic learning and may be less “visible” than other disorders, such as dyslexia and ADHD (e.g., Bishop et al., 2017; Christopulos & Kean, 2020; Norbury et al., 2016; Redmond, 2016; Tomblin et al., 1997).

Third, although we hypothesized that there would be stronger correlations between parent concerns and children's test scores in the online sample than the in-school sample, the results did not support this. Both samples exhibited similar correlations between parent concerns about literacy and language and reading scores, and the in-school sample showed stronger correlations between parent concerns about oral language and children's language scores. This was largely a function of the small number of parents reporting oral language concerns in the online sample. Taken together, although the samples were recruited by different means and differed in average caregiver education levels and financial resources (as measured by eligibility for free or reduced-price school lunch), the relationship between parent concerns and child abilities were generally similar across groups.

In considering these results, we acknowledge a few limitations and directions for future research. First, although a wide distribution of both language and reading abilities was observed in both samples, most children in both samples were performing within normal limits. We targeted recruitment efforts to reach children at risk for DLD and dyslexia (as well as typically developing children) for online study participation. However, this did not result in a higher proportion of children at risk for language or word reading problems than the in-school sample. Anecdotally, we observed that some parents who responded to our ads expressed concerns about their children's reading abilities but declined to enroll in the study after understanding that it was not an intervention study. A few parents of children in the online sample disclosed that they were teachers or SLPs, but unfortunately, we did not systematically collect data on parent occupations to examine this more closely. Overall, future studies with a larger number of children scoring in the below-average range will have more power to detect significant relationships between parent-reported concerns and child abilities.

Second, we must be cautious about any interpretations related to the history of support services within the in-school sample and between the in-school and online samples due to differences in the question in different years of the study. Third, while we aimed to address a broad range of concerns within the checklist, a potential limitation is that parents may have interpreted items differently than intended. For example, we intended for the item “Saying words correctly” to reference speech articulation skills. As pointed out by a reviewer, some parents may have considered pronunciation during oral reading when answering this question. The current study also focused on the relationship between language and literacy concerns with children's oral language and word reading scores. In addition to our single word reading measure, future work could examine children's text-level reading and/or reading comprehension skills. Future studies could also assess parent concerns about social language and include measures of pragmatic language skills. Finally, we acknowledge that binary responses about parent concerns may be less sensitive to the range of abilities as other types of parent reports, such as those that include a rating scale (e.g., Nelson et al., 2016), or ask parents to indicate the presence or absence of specific behaviors (e.g., Bishop, 2003). To this end, we are currently collecting data using a parent report measure that invites parents to rate their child's ability in various oral- and written-language domains.

Despite these limitations, we believe these findings have important implications for how children with language disorders are identified in research and in clinical practice. First, referral models that rely on parent and teacher concerns about language to prompt an evaluation may be contributing to the low rates of identification of children who meet criteria for DLD (Christopulos & Kean, 2020; Norbury et al., 2016; Tomblin, 2014). We found that most parents of children scoring at least 1 SD below the mean on oral language or reading responded “no” to each item on the questionnaire. Tomblin (2014) reported that the majority of children with DLD who were identified in an epidemiologic study in kindergarten and followed through high school did not receive intervention at any point. This is unfortunate, considering the high rate of reading comprehension difficulties observed in the same sample of children, especially in the later school grades (Catts et al., 2012), which may have been prevented or mitigated with effective language intervention. This finding aligns with studies showing that teachers are not able to identify markers of language impairment among school-age children (e.g., Antoniazzi et al., 2009; Christopulos & Kean, 2020). Secondly, the high comorbidity between language and other neurodevelopmental disorders including ADHD and dyslexia (Adlof & Hogan, 2018; Redmond, 2016; Snowling et al., 2020) suggests that when parents raise concerns about literacy, attention, or executive functions, an evaluation of oral language abilities is warranted. Although most parents did not report concerns, the percentage of parents reporting concerns for oral language and reading was higher for children who scored at least 1 SD below the mean in oral language or word reading than children who scored within normal limits. One possibility is that the observations that lead parents to be concerned about attention (e.g., not following directions) may be an indicator of language difficulties, but parents may not attribute those observations to oral language. Alternatively, language impairment may be a contributing source to the visible symptoms manifested in other domains.

Finally, our study speaks to the importance about raising public awareness about DLD. SLPs are experts in language development and disorders, and teachers and parents have knowledge about a child's language learning experiences in the school and at home. As a first step, we encourage SLPs and classroom teachers to build a shared understanding about language development and language disorders. This may include highlighting the reasons for why children with DLD are underidentified, the rates of co-occurrence of DLD with other disorders, discussing approaches for monitoring language development in the classroom, and understanding how teacher and parent awareness about language disorders and their associated risk factors can contribute to improving rates of identification. A checklist is one type of instrument that could be used to track concerns over time; however, this study does not provide strong evidence for it as a stand-alone tool. More importantly, the types of concerns that are being raised by parents/caregivers for children with low language and low reading scores may provide context about how and when language difficulties manifest in the home compared with a school setting and improve the information we gather from families. Improving the identification of children with DLD and increasing public awareness about language disorders requires a collaborative effort between clinicians, researchers, educators, and families.

Conclusions and Future Directions

This study is part of a larger longitudinal project about children with and without language and reading impairments. We found that parent concerns about oral language were low in school-age children, which may be attributed to lack of public awareness about language development and language disorders and the difficulty tracking language development without direct measurement. In raising public awareness about language development and language impairments specifically for DLD, we must consider methods that promote the identification of children with weaknesses. This study highlighted the relationship between parental concerns and children's scores on language and literacy. Further research is needed to examine the extent to which levels of parent awareness and parent concerns about language development can inform why identification rates vary in clinical and research settings for children with language-based disorders.

Tele-assessment has provided many benefits including broadening representation of learners in studies not bound by geographic location, but access to teletherapy services was not equal for students from lower versus higher socioeconomic backgrounds during the COVID-19 pandemic (Sylvan et al., 2020). Responses to our advertisements for online participants yielded a sample of participants with relatively highly educated parents with greater financial resources on average than our in-school sample. Still, the relationship between parent concerns and child abilities was largely similar across our two samples, despite their demographic differences. In our ongoing work, we aim to recruit a more racially, ethnically, and economically diverse sample of participants for online participation. Overall, we hope this research inspires researchers and clinicians alike to consider the influence of recruitment methods on understanding children with DLD in the broader population.

Author Contributions

Jessica Chan: Conceptualization (Supporting), Data curation (Lead), Formal analysis (Lead), Investigation (Supporting), Project administration (Supporting), Supervision (Supporting), Writing – original draft (Lead), Writing – review & editing (Equal). Suzanne M. Adlof: Conceptualization (Lead), Data curation (Supporting), Formal analysis (Supporting), Funding acquisition (Lead), Project administration (Supporting), Supervision (Supporting), Writing – original draft (Supporting), Writing – review & editing (Equal). Dawna Duff: Investigation (Supporting), Project administration (Supporting), Supervision (Supporting), Writing – original draft (Supporting), Writing – review & editing (Supporting). Alexis Mitchell: Investigation (Supporting), Project administration (Equal), Supervision (Equal), Writing – review & editing (Supporting). Maalavika Ragunathan: Investigation (Equal), Project administration (Supporting), Supervision (Equal). Anna M. Ehrhorn: Investigation (Supporting), Supervision (Supporting), Writing – review & editing (Supporting).

Funding Statement

Research reported in this article was supported by the National Institute on Deafness and Other Communication Disorders Grant R01DC017156, awarded to the University of South Carolina (PI: S. M. Adlof). The content is solely the responsibility of the author and does not necessarily represent the official views of the National Institutes of Health.