Sage Journals: Discover world-class research

Abstract

Despite the importance of receiving speech-language therapy for educational outcomes, the role of language in all subject learning, and parental language complexity as a hypothesized mechanism for reproducing educational inequalities, there is a surprising lack of attention to speech-language issues within sociology. The authors link theories from sociology of education and medical sociology and use the prospective longitudinal nationally representative UK Millennium Cohort Study to examine the roles that social determinants of health, teacher bias, and parental concern play in teacher recognition of speech-language needs in early childhood, net of a measure of speech-language ability. Although social determinants affect ability, there is little evidence of bias in teacher recognition once controlling for said ability, with the exception of lower recognition of female speech-language needs. However, social determinants also drive whether parents are concerned about their children’s speech-language ability, which has a striking association with teacher recognition for both under- and overperforming children.

Keywords

speech-language disorder educational inequality social determinants of health intergenerational inequality early childhood

Although a health issue, diagnosing and treating speech-language disorders is a critical part of the education system. Students with a speech-language disorder experience poorer educational, social, and vocational outcomes compared with students without these challenges (Elbro, Dalby, and Maarbjerg 2011; McGregor 2020). Fortunately, the sooner a child is enrolled in early intervention, the more likely they are to achieve successful outcomes (Guralnick 2011). The skills necessary for spoken language form the foundation for learning to read and write¹; however, these speech-language skills also form the basis for all subject learning. Students need to speak, converse, and read for all subjects, such that they can fall behind in all learning if they do not maintain age-appropriate speech-language skills. For sociologists, understanding factors that contribute to differences in speech-language ability is of additional importance because of their potential to reproduce inequality intergenerationally through spoken language (see Lareau 2003).

Given the importance of receiving speech-language therapy for educational outcomes, language’s role as the basis of all subject learning, and sociological theorizing that parental language complexity is a mechanism for reproducing inequality, the lack of attention to diagnosable and treatable speech-language issues among sociologists is notable. Among generalist journals and relevant American Sociological Association subfield journals, only two articles examined issues related to speech-language disorders (Hibel, Farkas, and Morgan 2010; Skrtic, Saatcioglu, and Nichols 2021).² Both articles consider speech-language impairment as one outcome among several disability statuses, with overrepresentation among non-Whites and boys. As we next describe, there are numerous additional social determinants associated with receiving a formal diagnosis of a speech-language disorder. Informed by work in sociology of education, medical sociology, and the field of speech-language pathology, we assess the role social determinants, teacher bias, and parental concern play in teacher recognition of speech-language needs in early childhood.

Social Determinants of Speech-Language Disorders and Schools as Equalizers

A number of social determinants are known risk factors for child speech-language disorders. In the United States, an estimated 3 percent to 16 percent of children have speech-language disorders (National Academies of Sciences, Engineering, and Medicine 2016). In the United Kingdom, 10 percent of children are estimated to have long-term speech-language needs, which rises to more than 50 percent in areas of high deprivation (Public Health England 2020). Despite this, only about 3 percent of UK pupils aged 7 have been identified as having speech, language, and communication needs (Meschi et al. 2010). In a U.S.-based study using a random sample to establish prevalence rates (7.4 percent), only 29 percent of parents of children who met criteria for speech-language impairment reported being previously informed that their children had speech or language problems (Tomblin et al. 1997).

With such discrepancies for both schools and parents, a natural question is whether there is any patterning to which children are recognized as in need of treatment. Compared with girls, boys are more likely to have a speech-language disorder (Campbell et al. 2003; Tomblin et al. 1997) and disproportionately be identified and added to clinical caseloads (Zhang and Tomblin 2000). Differences in social behavior act to magnify speech-language difficulties in boys but hide these difficulties in girls (Hart et al. 2004; Toseeb et al. 2017). Socioeconomic disadvantage also increases the risk for having a speech-language disorder (Roy, Chiat, and Dodd 2014). A number of other risk factors are well documented in the literature, including family history of speech-language disorder, perinatal complications (premature birth, low birth weight, neonatal intensive care stay), and persistent middle-ear infections (Campbell et al. 2003). Much of this literature fits within the core conjectures of the theory of fundamental causes from medical sociology. According to this theory, socioeconomic differences in health outcomes are entrenched across societies. The reason for this consistency is that those at the lower end of the socioeconomic status (SES) spectrum lack the power, knowledge, and resources to take advantage of existing and novel health interventions (e.g., Chang and Lauderdale 2009; Glied and Lleras-Muney 2008; Link and Phelan 1995; Masters, Link, and Phelan 2015; Polonijo and Carpiano 2013), a tenet that will become important below. Figure 1 introduces our theoretical model, with the role of social determinants reflected in path A.

Figure 1.

Theoretical model.

As shown in path B in Figure 1, school and health interventions might possess the ability to act as equalizers, bypassing and potentially leveling the preexisting inequalities in path A. We consider this to be the most objective pathway by which schools and teachers can operate: school and teacher recognition of a particular student need should be driven primarily by whether the assessment identifies a particular issue. By contrast, recognition should not be driven by social determinants associated with assessment performance or even those unrelated to performance but nonetheless based on discriminatory bias. Importantly, this mechanism could allow schools to act as equalizers; by identifying students in need of intervention, schools could level the playing field regarding the inequalities associated with assessment scores. According to research on schools as equalizers, the largest variance in achievement by SES is at the start of formal schooling, which shrinks afterward (von Hippel, Workman, and Downey 2018). Moreover, differences by race/ethnicity, class, and gender tend to expand when school is out of session and contract during the school year or are nonexistent for those who attend year round (Condron, Downey, and Kuhfeld 2021; Downey 2020; Downey, Kuhfeld, and van Hek 2022; Downey, von Hippel, and Broh 2004; Johnson and Wagner 2017). The degree to which the racial/ethnic and SES composition of schools affects such equalization, however, is a matter of contention (Condron 2009; Downey, Quinn, and Alcaraz 2019).

In a similar vein, medical sociology has shown that such bypassing of inequality via “upstream intervention” is particularly likely when the intervention applies equally across SES; that is, when power, knowledge, and resources are not required to take advantage of the intervention (e.g., Dowd and Aiello 2008; Hernandez 2013; Hernandez et al. 2019; Polonijo 2020). Such a case may be true of a universal assessment within schools to identify health-related issues and particularly for those health issues closely related to education itself, such as speech-language disorders. Indeed, the education system is considered a primary upstream point of intervention to improve health-related outcomes, as it is here that knowledge can be generated and intergenerational inequality addressed (Hummer and Hernandez 2013; Zajacova and Lawrence 2018). Thus, schools have the potential to reduce inequality when they are tasked with assessing children for health issues. If schools can identify children with speech-language needs and address those needs, then they could bring such students up to the skills of their peers and potentially eliminate inequality associated with sources of (dis)advantage in need identification.

Although with varying levels of success and not without controversy, schools have been used as a mechanism for reducing inequality in health. Many such policies use schools as a way to reach children such that health interventions are applied nearly universally but have little to do directly with educational outcomes. Examples include body mass index screenings (Ikeda, Crawford, and Woodward-Lopez 2006), vaccine mandates (Polonijo 2020), and scoliosis tests (Yawn et al. 1999). Other screens are conducted within schools to identify health issues that directly affect learning, including hearing, vision, and speech-language difficulties. Screening for speech-language difficulties is unique in several ways. First, when a vision or hearing screening or any health intervention unrelated to educational outcomes indicate a child has needs, the school is not the institution that addresses that need. Rather, the result is passed to the parents for interfacing with the health care system. By contrast, when speech-language needs are identified for a child, the onus is often on the school system to provide therapy through a speech-language pathologist. Second, vision and hearing screenings are typically conducted by specialists. Although some jurisdictions have universal speech-language screenings conducted by speech-language pathologists, some screens are conducted by teachers, including the one that we introduce below for England. Other than this assessment, the classroom teacher becomes the primary actor within schools for identifying speech-language needs. Despite schools’ potential equalizing role, there is another body of research that argues schools and teachers may actually perpetuate inequality.

Schools and Teachers as Perpetuating Inequality

Contrasting with schools as potential equalizers, there is also considerable evidence that existing inequalities are perpetuated by schools and teachers, represented in path C in Figure 1. Some portion of inequality occurs at the school level, whether between schools, through how residential segregation is associated with socioeconomic and demographic characteristics of the students (e.g., Condron 2009; Hanselman and Fiel 2017; Hibel et al. 2010), or within schools, such as through how students are sorted into classrooms (e.g., Condron 2007; Lewis and Diamond 2015). Alongside these macro and meso levels, the interactions that occur in the classroom between teachers and students also contribute to perpetuating inequality.

A considerable body of research demonstrates that teachers perceive and treat students of varying race, gender, and class backgrounds differently, which can affect learning, achievement, and recognition of need. Although the theory of an “oppositional culture” to education among certain racial/ethnic minority groups has largely been proved false, it has had a lasting effect on teachers’ perceptions, whereby they view racial/ethnic minorities, especially Black students, as having lower academic orientations even in the face of comparable academic records (Downey and Pribesh 2004; Hanselman 2018; Harris 2011; Lewis and Diamond 2015; McGrady and Reynolds 2013; Tyson 2011). It is important to note that speech-language patterns associated with dialect or learning English as a second language do affect teacher perceptions (Canizales 2021; Carter 2006; Holland 2012; McGrady and Reynolds 2013; Morris 2005a, 2005b) but are unequivocally not speech-language disorders and should not be used as the basis for evaluating speech-language skills (as opposed to the [in]ability to, e.g., accurately produce the sounds and sentence structures expected for a child’s age in that language or dialect) (ASHA 2003). Yet racial/ethnic differences in teacher perceptions of the related area of literacy skills remain even after accounting for an objective measure of such skills (Irizarry 2015). Perversely, students whose literacy skills were overestimated by teachers relative to objective measures, largely on the basis of sociodemographics, gain more literacy skills during kindergarten (Ready and Chu 2015).

Complicating these racial/ethnic perceptions are interactions with gender and class. Relative to White boys, Black and Latino boys are viewed by teachers as more blameworthy for identical, routine classroom misbehavior (Owens 2022) and thus experience disciplinary and corrective action more often (Gansen 2021; Morris 2005a, 2005b, 2007). Individual-level personality characteristics are also perceived differently by teachers; for example, unlike other identities for which assertiveness was rewarded, Black girls’ assertiveness was viewed as abrasive or aggressive, pushing them toward passivity and silence in the classroom (Morris 2007). Teacher perceptions across race and gender are also influenced by students’ social class positions (Morris 2005a; Musto 2019). Teacher treatment subsequently reifies elementary students’ self-perceptions and reproduces inequalities by race, class, and gender (Harvey 2023).

These teacher effects related to race, gender, and class can potentially be consequential for identifying speech-language needs. If teachers are outright biased, perceive language differences as part of an oppositional culture, misperceive the role of English as a second language, or have differing views of individual blameworthiness, they may be less likely to attribute clinically meaningful speech-language issues to a real need for services for marginalized groups. However, there is one additional important stakeholder to consider: parents.

The Role of Parents and Inequality

Among the most well-known concepts regarding the role parents play in perpetuating inequality in schooling is Lareau’s (2003) theory of concerted cultivation, largely described as a dichotomy of parenting strategies based on the cultural capital available to higher class families relative to lower class families. Concerted cultivation highlights how higher class parents use demanding patterns of language, interact more with institutions in which their children are involved, and emphasize highly structured activities outside school. Taking these factors in turn, parents speak differently to their children, with higher class parents using more cognitively demanding modes of speech and reasoning and negotiating with their children relative to lower class parents’ use of directives and restricted codes of speech. As noted earlier, research in speech-language development also provides empirical evidence of class differences (Hoff 2003; Roy et al. 2014). Thus, higher class parents may be especially keen to address perceived speech-language concerns, whereas lower class parents may not possess the same knowledge of the importance of speech-language development.

Sociologists have focused especially on the parental interaction with institutions component of concerted cultivation. According to Lareau (2003), higher class parents are more comfortable interacting with professionals such as physicians and educators. Thus, they are more likely to seek out interactions with teachers and more involved with their children’s schooling. Even when middle-class parents choose to send children to urban schools, their involvement, although providing some benefits, can also unintentionally marginalize less affluent parents, subsequently diminishing lower class families’ participation in improving schools (Posey-Maddox 2014). Parents further influence the degree to which teacher bias may play out, as teachers are reluctant to confront higher SES White parents who are disproportionately involved in school administration and more likely to challenge decisions (Calarco 2020; Lewis and Diamond 2015). For speech-language needs, teachers may be reluctant to confront or contradict such parents who argue that their children require additional services even when they do not.

The influence of parents, however, also operates through mechanisms that do not require direct involvement in schooling. Although Lareau (2003) emphasized structured activity, parents and the home also act as stage setting for education through whether the environment is conducive to learning and the message that parents convey about education (Harris and Robinson 2016). That is, nearly all parents express value in education (Harris and Robinson 2016), but the resources they are able to draw on confer advantages upon the already privileged (Buchmann, Condron, and Roscigno 2010). One study in England supported these findings for the advantaged but emphasized that working-class parents engaged in “essential assistance,” providing what they could to help their children stay on track and out of trouble (Wheeler 2018). Such stage setting occurs in studies of children’s role in inequality in the classroom, with middle-class elementary students more proactive in seeking help from teachers relative to working-class students (Calarco 2011). Thus, advantaged parents who express concern about their children’s speech-language needs might not only directly advocate for their children to teachers but also influence the degree to which their children act to receive related help from teachers.

This potential influence of parents is shown through paths D and E in Figure 1. Bridging medical sociology and sociology of education through speech-language needs, concerted cultivation maps onto the three main concepts from fundamental causes theory. As shaped by social determinants in path D, parents require the knowledge to recognize potential speech-language needs, the power to feel that they can intervene within schools, and the resources to provide a stage-setting environment conducive to learning and therapy. Ultimately, these parental processes may affect the degree to which teachers recognize a child’s speech-language needs via path E. In what follows, we piece together each of the described pathways to assess the role social determinants, teacher bias, and parental concern play in teacher recognition, net of a measure of speech-language ability.

Methods

Data

We used the intergenerational, prospective UK Millennium Cohort Study (MCS), a longitudinal and nationally representative sample of infants born from September 2000 to January 2002 (referred to as “2001”).³ The survey has conducted eight waves, also known as sweeps, with the most recent in 2023 at age 23. Parents took surveys in all sweeps, with children self-responding in later sweeps. We used data from the first five sweeps: sweep 1 at age 9 months (2001), sweep 2 at age 3 (2004), sweep 3 at age 5 (2006), sweep 4 at age 7 (2008), and (in supplemental analyses) sweep 5 at age 11 (2012). The first sweep included 18,552 families. An additional 692 families were included in sweep 2 after updating the sample registry of eligible births. For such families, we used sweep 2 information for the background characteristics described below. For simplicity, we refer to a family’s “baseline interview” when describing this information. In sweep 4, 13,857 families were still participating. Stratified sampling was used by country and then on the basis of ward characteristics to oversample economically disadvantaged wards and, in England, wards with high percentages of non-White population. The survey administrators provided weights that adjust for both attrition and sampling design. All analyses used these weights via the svy suite of commands in Stata.

Over the eight sweeps, the MCS conducted numerous ancillary surveys and provided linked external data. We used three additional MCS data sources of information. First, we used the linked dataset for the participating children’s foundation stage profile (FSP). Following ages 3 to 5, known as the “early years foundations stage,” the FSP is conducted countrywide in England annually as part of the standard curriculum to assess achievement of specific learning goals at age 5 before entry into what is formally known as year 1 of schooling. According to the UK’s Special Educational Needs (SEN) Code of Practice concerning the joint responsibilities of the education and health system, the FSP “should inform plans for future learning and identify any additional needs for support.” There are six areas of learning covering children’s physical, intellectual, emotional, and social development. Across these six areas, there are 13 assessment scales conducted by the child’s teacher specifically trained in assessment administration. The area we used is “communication, language and literacy,” specifically the “linking sounds and letters” (LSL) scale, described further in the next section. FSP results are passed on to parents and the child’s next teacher. MCS administrators had a 95 percent successful matching rate with the official FSP dataset in England.

There is no equivalent to the FSP conducted in Wales, Scotland, or Northern Ireland. Thus, our second additional information source is the sweep 3 (thus also age 5) Celtic Country Teacher Survey (CCTS). A questionnaire was sent to teachers of MCS cohort children attending schools in those three countries to replicate the information collected by the FSP. The CCTS teacher response rate was 54 percent in Wales, 59 percent in Scotland, and 68 percent in Northern Ireland. We note that all results presented were similar if restricted to the official linked FSP data in England.

Finally, we used the teacher survey conducted with sweep 4 at age 7, which, aside from the limited CCTS, was the only teacher survey conducted. Regardless of country, survey administrators attempted to contact all teachers of cohort members. Teachers of 8,876 cohort members participated, with similar response rates across countries.

Variables

FSP Speech-Language Score

As noted earlier, we used the LSL FSP scale, as this scale is designed to assess progress in speech-language development. We note that this scale has been used in other studies in the field of speech-language pathology (e.g., Forrest et al. 2020; St. Clair et al. 2019). All FSP scales have a similar nine-point progression for behind, on-track, and ahead of learning goals. The first three LSL points represent prerequisite skills that are expected to be in place by age 5. These skills are (1) joins in with rhyming and rhythmic activities, (2) shows an awareness of rhyme and alliteration, and (3) links some sounds to letters. Once these skills are achieved, the next five points represent the actual LSL learning goals and progress in order of difficulty, although they can be achieved in any order. They are (4) links sounds to letters, naming and sounding letters of the alphabet; (5) hears and says sounds in words; (6) blends sounds in words; (7) uses phonic knowledge to read simple regular words; and (8) attempts to read more complex words, using phonic knowledge. Upon achieving these eight goals, the final point indicating achievement beyond the LSL early learning goals is (9) uses knowledge of letters, sounds, and words when reading and writing independently.

In our models, we treated the FSP as a continuous measure ranging from 0 to 9. We also show descriptive and bivariate relationships across the 10 score categories and discuss the robustness of our models to categorical coding. Furthermore, as speech-language and other health assessments often use standardized scores to determine whether a child is behind in development, we also discuss the standardized version of the scale.

Age 5 and Age 3 Parental Speech-Language Concern

In sweeps 2 and 3, parents were asked, “Do you have any concerns about [child’s] speech and language?” The question was specifically intended not to include concerns about learning English as another language, with interviewers told to say, “This question does not include concerns about learning English.” We coded 1 for concerned and 0 for not concerned. We primarily used the age 5 measure to be able to both control for the age 5 FSP LSL score in models for parental concern and to place parental concern in the sweep immediately before the age 7 teacher survey. However, we used the age 3 measure in models for the age 5 FSP LSL score to establish temporality.

Age 7 Speech-Language SEN

In sweep 4, parents indicated whether the child has received a SEN from the school or local education authority specifically for a “problem with speech or language.” This measure was first collected at age 7, limiting its inclusion to the final model assessing teacher recognition that is also at that age. We also considered this outcome at age 11 in supplementary models.

Age 7 Teacher Recognition of Speech-Language Needs

In the sweep 4 teacher survey, teachers were asked, “Do these specific problem(s) apply to this child?” referring to the participating cohort member. Among a series of choices was “problem with speech or language.” We coded 1 for yes and 0 for no. We note that teachers can (and do) select yes regardless of whether the child has a SEN and thus receives additional support.

Baseline Background Characteristics

We included numerous background characteristics collected at baseline interview indicative of social determinants and additional controls. To assess class, we included whether the parent had the equivalent of a bachelor’s degree or higher (no as reference) and family income quintiles (first as reference). We also included biological sex (male as reference) and an indicator for race as White (non-White as reference).

Various household characteristics included whether another language was spoken at home (no as reference), the number of parents in the household (single parent as reference), and the number of siblings. Measures related to the child’s birth included the responding parent’s age at birth, the child’s birth weight, and whether the child was “taken to the special care or neo-natal or intensive care unit after birth,” which we refer to as “special care at birth” for short.

As neurodevelopmental disorders tend to run in families, we included an indicator for whether the parent had a long-standing mental, behavioral, or neurodevelopmental (MBN) disorder. Parents were asked, “Do you have a longstanding illness, disability, or infirmary?” If yes, interviewers asked, “What is the matter with you?” Survey administrators then coded and provided the answer on the basis of the International Classification of Diseases, Tenth Edition. Chapter F, “Mental, Behavioral, and Neurodevelopmental Disorders,” contains developmental speech-language disorders as well as many of the disorders that are comorbid with them.

At the aggregate level, we included the health deprivation and disability decile, collected and released by the official statistics agencies in each UK country. These measures were computed at the local authority district level and then ranked within each of the four countries (e.g., there were 32,842 districts in England at baseline). Although not precisely equivalent across countries, they are broadly comparable when using deciles, according to MCS user guides. The indicators within this domain included years of potential life lost, comparative illness and disability ratio, acute morbidity, and mood and anxiety disorders. Higher deciles indicate higher deprivation.

Finally, we included a control for country (England as reference). We caution that the dummy variables may not represent true country differences, as the idiosyncrasies of differences across surveys and location (e.g., FSP vs. CCTS, country-specific questions in the sweep 4 teacher survey, indigenous Celtic language use outside of England) could be the reason for differences. Thus, we view the inclusion of country (and the analytic weight adjustments) as assuaging concerns about these differences rather than reflecting true differences across country.

Analysis

We present models for three outcomes. First, we use linear regression for the age 5 FSP LSL score to determine which background characteristics were associated with the objective speech-language assessment. Second, we present logistic regressions of age 5 parental speech-language concerns to determine, controlling for the FSP objective assessment, how background characteristics affected who was concerned about their children’s speech and language. We present a model for children scoring three or below, as these scores indicate a child who is behind age-appropriate learning goals. We also present a model for children scoring seven or higher to examine if certain parents expressed concern even though the assessment indicated their child was progressing satisfactorily. Third, we show logistic regressions of age 7 teacher recognition of speech-language needs, with particular interest in how the objective FSP assessment and parental concern affected this recognition, net of background characteristics.

With the exception of including the FSP LSL score as a control in the parental speech-language concerns model (both age 5) and student SEN status in the teacher recognition model (both age 7), all predictors came from the prior sweep or baseline survey to establish temporal order (e.g., parents may be more concerned if the teacher recognizes a speech-language problem). Furthermore, separating the age 5 FSP from the age 7 teacher evaluation ensured that it was extremely unlikely that the same teacher who performed the assessment was the child’s teacher two years later. As noted earlier, all analyses included weights for attrition and survey design. Across models, the maximum average variance inflation factor of 1.52 was well below the cutoff for multicollinearity.

For each model and descriptive statistics, we sought to retain the maximum possible sample size. There were virtually no missing data on baseline background characteristics. Sample size was reduced, however, when using data from adjacent surveys (i.e., sweep 2 variables in sweep 3 models), as participants can come in and out of the survey. Our sample size was further reduced in the sweep 4 teacher survey models because of analogous adjacent sweep attrition, the response rate on the teacher survey, and the CCTS response rate. Descriptive statistics are shown in Table 1. As can be seen, the characteristics of the reduced sample size in the sweep 4 analytic sample were nearly identical to that of the sweep 3 analytic sample.

Table 1.

Descriptive Statistics.

	Sweep 3 Data (n = 11,761)		Sweep 4 Data (n = 6,498)
	Mean	SE	Mean	SE
Age 7 (sweep 4)
Teacher recognition of child speech-language needs			6.7%
School special education need for speech-language			1.8%
Age 5 (sweep 3)
Parent speech-language concern (age 5)	13.1%		12.3%
Foundation stage profile linking sounds and letters score	6.262	.040	6.442	.043
Age 9 months (sweep 1)^a
Country
England	87.0%		85.5%
Wales	3.2%		3.3%
Scotland	6.5%		7.5%
Northern Ireland	3.3%		3.7%
Income Quintile
First	20.5%		18.6%
Second	19.9%		19.8%
Third	20.6%		20.8%
Fourth	20.3%		21.3%
Fifth	18.7%		19.6%
Parent bachelor’s degree or higher	16.8%		17.3%
Race
White	87.0%		88.6%
Non-White	13.0%		11.4%
Sex
Male	50.7%		49.4%
Female	49.3%		50.6%
Another language spoken at home	10.5%		9.4%
Parent age at birth (years)	28.744	.123	28.765	.139
Number of parents in household
Two parents	82.9%		84.1%
Single parent	17.1%		15.9%
Number of siblings	1.135	.013	1.122	.015
Parent long-standing mental, behavioral, or neurodevelopmental disorder	4.1%		4.3%
Birth weight (kg)	3.350	.007	3.367	.009
Special care at birth	9.2%		8.6%
Aggregate health deprivation and disability decile	5.536	.121	5.389	.123

Note: Weighted for complex survey design and attrition.

A small number of eligible families started in sweep 2; data for variables labeled sweep 1 come from sweep 2 (age 3) for these families. See the “Data” section for more information.

Results

Descriptive Statistics

Given the similarity between the sweep 3 and 4 analytic samples in Table 1, we describe sweep 4 for illustrative purposes. At age 7, teachers stated that 6.7 percent of MCS children had speech-language problems, while 1.8 percent of parents reported a school SEN for speech-language needs. Among parents, 12.3 percent were concerned about their children’s speech-language skills. On the FSP LSL (range = 0–9), the average score was 6.4. Table 2 shows percentages across the 10 scores. Most children performed in the range of 4 to 8, indicating that they were meeting the learning goals, with 7 the most common score at 19.1 percent. About 14.7 percent of children scored 3 or lower, indicating that they had not progressed to the actual learning goals for a student aged 5. Such students were at least 1.44 standard deviations below the average score. With a score of 9, 18.3 percent were performing above expectations.

Table 2.

FSP Score (Age 5) by Parental SL Concerns (Ages 3 and 5) and Teacher and School Recognition of Child SL Needs (Age 7).

Descriptives for Foundation Stage Profile Scores			Percentages “Yes” by Foundation Stage Profile Score				Percentages for Teacher SL Needs Age 7 by FSP Score and Parental Concern
FSP Score Age 5	Standardized FSP Score Age 5	Percentage	Parental SL Concern Age 3	Parental SL Concern Age 5	Teacher SL Needs Age 7	School SEN for SL Age 7	No Parental SL Concern Age 5	Parental SL Concern Age 5
0	−2.76	.6%	60.9%	65.7%	65.0%	23.0%	28.2%	98.0%
1	−2.32	1.5%	40.6%	47.9%	52.7%	16.6%	27.4%	77.7%
2	−1.99	3.5%	25.3%	29.5%	31.9%	9.2%	18.2%	67.4%
3	−1.44	9.1%	20.9%	25.1%	20.1%	3.8%	10.0%	50.0%
4	−1.00	9.0%	15.7%	17.5%	11.4%	3.6%	7.3%	32.0%
5	−.56	10.5%	15.1%	15.8%	8.1%	2.2%	4.3%	29.8%
6	−.12	12.3%	13.5%	11.9%	4.9%	.9%	2.1%	23.9%
7	.32	19.1%	12.1%	9.5%	3.4%	.6%	1.9%	18.0%
8	.76	16.1%	9.9%	7.8%	1.2%	.2%	.5%	9.5%
9	1.20	18.3%	8.0%	5.6%	.6%	.4%	.3%	6.1%
n = 11,989			n = 10,832	n = 11,988	n = 6,564	n = 10,473	n = 5,757	n = 807

Note: Weighted for complex survey design and attrition. FSP = foundation stage profile; SEN = special education need; SL = speech-language.

Table 2 examines the relationship among our three focal variables of the FSP LSL score, parental speech-language concerns, and teacher recognition of speech-language needs. The middle columns show the percentage of parents concerned about their children’s speech and language and teachers answering that the child had speech-language needs across the 10 FSP scores. Although the overall pattern was as expected, in which there was a higher percentage of concerned parents and teacher recognition at lower scores, the percentages did not fully correspond to the assessment results. That is, at scores of 3 or lower, parental concern is justified, as such children were assessed as not having progressed to speech-language learning goals for their age. But even at score 0, only 65.7 percent of parents expressed concern at the coterminous age 5 data collection. At scores 1, 2, and 3, the percentages of parents concerned were 47.9 percent, 29.5 percent, and 25.1 percent, respectively. At the high end of the distribution, there were also concerned parents whose children was performing well. Indeed, 5.6 percent of parents whose children were performing ahead of expectations nonetheless expressed concern. The teacher percentages were more concerning. Children who were assessed to be performing below expectations at age 5 were in many instances not recognized by the teacher as having speech-language needs at age 7. For example, for scores 0 and 1, teachers expressed a problem with speech-language for only 65.0 percent and 52.7 percent of such children, respectively. As striking, a low percentage of children performing below expectations have an official school SEN at age 7, with 23.0 percent, 16.6 percent, and 9.2 percent for scores of 0 through 2, respectively. Given these differences, teacher recognition is not solely accounted for by a school SEN: among students with a SEN, teachers only recognize a speech-language need for 77.2 percent of such students; for those without a SEN, teachers state that 5.7 percent of such students have a speech-language problem.

Although the two-year gap may explain some of the teacher discrepancy, the third set of columns show that what may be more likely at play was a relationship between parental concern and whether teachers recognize speech-language needs. At the lowest score, if parents were concerned about their children’s speech-language skills, the child’s needs were virtually guaranteed to be recognized by the teacher at 98.0 percent, compared with teacher recognition in only 28.2 percent of those children whose parents did not express concern. Although a score of 0 was uncommon, these differences can be seen for all children performing below age 5 expectations with scores 3 or lower. Although some children might have progressed since age 5, these differences are stark given that many of the children in this score range were almost certainly still in need of speech-language services given how far behind they were at age 5 (Hayiou-Thomas et al. 2010) and language disorders tend to persist into adulthood if not resolved by this age (McGregor 2020). Gaps by parental concern occurred across the entire range of the FSP LSL score. Surprisingly, teacher recognition still differed by whether there was parental concern for children who scored the highest and were progressing satisfactorily; at a score of 9, teachers recognized need among 6.1 percent of children if parents expressed concern, but teacher recognition was only 0.3 percent if parents did not. These descriptive relationships provide an initial indication that parental recognition of speech-language needs matters, as teacher recognition of such needs differed dramatically, even among students who should objectively need (or not need) services on the basis of their FSP assessments. We next turn to regression models to elaborate these relationships further.

Regression of FSP Speech-Language Score at Age 5

As a starting point, we present a linear regression of the children’s FSP LSL score at age 5 by baseline characteristics, shown in Table 3, model 1. Although there were no race differences, girls scored on average 0.556 points higher than boys (p < .001), and higher SES was associated with higher scores. Children of parents with bachelor’s degrees scored 0.633 points higher than those whose parents had lower educational attainment (p < .001), and, for example, those in the highest income quintile scored almost a full point higher (b = 0.945, p < .001) than those in the lowest quintile. Single-parent households (b = −0.227, p < .01) and a greater number of siblings (b = −0.280, p < .001) were associated with lower scores. Surprisingly, given prior literature, speaking another language at home and parent long-standing MBN disorders were not significant. Birth characteristics also were relevant, with each increase in parental age associated with 0.025 higher scores (p < .001), each increase in kilogram weight at birth associated with 0.243 higher scores (p < .001), and those children who required special care at birth scoring 0.250 points lower than those not requiring care (p < .01). Finally, each one-unit increase in the aggregate health deprivation and disability scale was associated with 0.026 fewer points (p < .05). In model 2, we included parental speech-language concerns at age 3 to show that the associations between these background characteristics and the assessment score are robust to its inclusion. Children of concerned parents scored 0.743 points lower on average (p < .001).

Table 3.

Linear Regression of Foundation Stage Profile SL Score at Age 5.

	Model 1	Model 2
Country (vs. England)
Wales	.411** (.140)	.395** (.139)
Scotland	1.508*** (.084)	1.499*** (.084)
Northern Ireland	.333*** (.098)	.356*** (.096)
Income quintile (vs. first)
Second	.340*** (.086)	.318*** (.084)
Third	.755*** (.088)	.740*** (.087)
Fourth	.908*** (.096)	.878*** (.094)
Fifth	.945*** (.094)	.906*** (.093)
Parent bachelor’s degree or higher	.633*** (.055)	.633*** (.054)
Race: White (vs. non-White)	.007 (.106)	.020 (.103)
Sex: female (vs. male)	.556*** (.047)	.495*** (.046)
Another language spoken at home (vs. no)	−.169 (.098)	−.175 (.099)
Parent age at birth	.025*** (.005)	.024*** (.005)
Single-parent household (vs. two)	−.227** (.075)	−.214** (.075)
Number of siblings	−.280*** (.025)	−.274*** (.025)
Parent long-standing MBN disorder (vs. no)	−.098 (.102)	−.059 (.101)
Birth weight	.243*** (.046)	.237*** (.045)
Special care at birth (vs. no)	−.250** (.080)	−.240** (.079)
Aggregate health deprivation and disability decile	−.026* (.012)	−.026* (.012)
Parental SL concern (age 3)		−.743*** (.069)
Constant	3.930*** (.231)	4.097*** (.226)

Note: Values in parentheses are standard errors. Weighted for complex survey design and attrition. N = 10,648. MBN = mental, behavioral, and neurodevelopmental; SL = speech-language.

p < .05. **p < .01. ***p < .001.

Many of these associations support known associations in the field of speech-language pathology. Although these relationships support fundamental causes theory, as the most disadvantaged children scored lower on the assessment, schools could be an equalizer by referring children to appropriate services to address this health and learning issue. By contrast, fundamental causes could still play a role in two ways. First, teachers could use this background information beyond its association with assessment scores in ways that exacerbate disadvantage. Second, teacher recognition of speech-language needs was related to parental ability to recognize speech-language needs in the descriptive statistics in Table 2, and this ability may be associated with (dis)advantage in a manner that exacerbates inequality. Next, we consider this intermediate step of parental concern before considering teachers.

Logistic Regression of Parental Speech-Language Concern at Age 5

In model 3 in Table 4, we examine factors affecting parental speech-language concerns among children who scored 3 or lower on the FSP speech-language assessment. Recall that for these children, parental concern is justified as they have not met age-appropriate goals. As expected, FSP scores in this range were associated with higher concern. Among the other measures (aside from country), only two predictors were significant. First, having a female child was associated with 33.6 percent lower odds of parental concern relative to male children (p < .01). Thus, although girls scored higher on the FSP scale, girls most in need of speech-language services were less likely to have concerned parents. This relationship could potentially translate into parents not advocating as strongly on behalf of their female children who need services. Second, being a parent with a long-standing MBN disorder was associated with 2.5 times higher odds of concern relative to parents without such disorders (p < .01). Interestingly, these disorders were not associated with objective assessment scores in Table 3, but parents perhaps were concerned that their children were at risk for a similar disorder to themselves.

Table 4.

Logistic Regression of Parental Speech-Language Concerns at Age 5 for Children Scoring 3 or Less and 7 or Higher on the Foundation Stage Profile Speech-Language Score.

	Model 3: FSP LSL Score 3 or Lower	Model 4: FSP LSL Score 7 or Higher
FSP LSL score	.616*** (.048)	.746*** (.055)
Country (vs. England)
Wales	.586** (.101)	1.011 (.171)
Scotland	1.578 (.506)	1.433* (.218)
Northern Ireland	.832 (.187)	1.458* (.213)
Income quintile (vs. first)
Second	.904 (.142)	.803 (.170)
Third	1.266 (.286)	.924 (.190)
Fourth	1.015 (.265)	.666 (.172)
Fifth	.757 (.250)	.631 (.151)
Parent bachelor’s degree or higher	1.211 (.347)	1.390** (.174)
Race: White (vs. non-White)	1.048 (.240)	2.323** (.595)
Sex: female (vs. male)	.664** (.089)	.587*** (.061)
Another language spoken at home (vs. no)	.626 (.158)	1.300 (.305)
Parent age at birth	.994 (.012)	.994 (.013)
Single-parent household (vs. two)	.787 (.135)	.657* (.139)
Number of siblings	1.027 (.063)	1.193** (.069)
Parent long-standing MBN disorder (vs. no)	2.453** (.687)	1.444 (.316)
Birth weight	1.040 (.107)	.922 (.095)
Special Care at birth (vs. no)	1.235 (.255)	1.172 (.210)
Aggregate health deprivation and disability decile	1.016 (.024)	.972 (.022)

Note: Exponentiated coefficients with standard errors in parentheses. Weighted for complex survey design and attrition. N = 1,769 in model 3, n = 6,355 in model 4. FSP = foundation stage profile; LSL = linking sounds and letters; MBN = mental, behavioral, and neurodevelopmental.

p < .05. **p < .01. ***p < .001.

In model 4 in Table 4, we consider parental concern among the subset of children where there was likely little reason for concern by examining those at the high end of the assessment scores using 7 through 9. Race, class, and gender associations emerged among this subset. The associated odds of speech-language concerns were 39.0 percent higher among parents with a bachelor’s degree than parents with lower education levels (p < .01), 2.3 times higher among White parents than non-White parents (p < .01), and 41.3 percent lower among parents of girls than boys (p < .001). Thus, advantaged status group membership may allow parents to express concern for a child that presumably does not need speech-language services. Children in two-parent households and those with more siblings also had higher odds of parental concern. Next, we consider how parental concern contributes to teacher recognition of speech-language needs.

Logistic Regression of Teacher Recognition of Speech-Language Needs at Age 7

Table 5 shows logistic regression models predicting teacher recognition of speech-language needs at age 7. Model 5 includes the same background characteristics that were used to predict the age 5 FSP LSL score. Without considering assessment scores, we would expect the same background characteristics to drive teacher recognition. Indeed, this was largely the case, with negative associations with income, parent’s education, female sex, and birth weight, and a positive association with number of siblings. Although not significant in predicting the FSP score, parental MBN disorder was associated with 65.4 percent higher odds of teacher recognition (p < .05).

Table 5.

Logistic Regression of Teacher Recognition of Child SL Needs at Age 7.

	Model 5	Model 6	Model 7	Model 8	Model 9	Model 10
Foundation stage profile LSL score		.558*** (.016)	.585*** (.018)	.597*** (.018)	.621*** (.020)	.639*** (.022)
Parental speech-language concern				10.048*** (1.274)	9.749*** (1.321)	7.762*** (1.132)
Country (vs. England)
Wales	1.093 (.193)		1.178 (.254)		1.338 (.263)	1.312 (.254)
Scotland	.641 (.160)		1.412 (.425)		1.266 (.406)	1.216 (.406)
Northern Ireland	.741 (.179)		.639 (.168)		.550* (.142)	.583* (.159)
Income quintile (vs. first)
Second	.524*** (.087)		.592** (.100)		.635* (.125)	.649* (.131)
Third	.450*** (.093)		.671 (.145)		.661 (.156)	.597* (.142)
Fourth	.352*** (.077)		.523** (.127)		.582* (.157)	.482* (.147)
Fifth	.349*** (.088)		.578* (.154)		.659 (.189)	.575 (.173)
Parent bachelor’s degree or higher	.450*** (.101)		.659 (.153)		.639 (.154)	.653 (.163)
Race: White (vs. non-White)	1.260 (.362)		1.337 (.386)		1.142 (.341)	1.206 (.388)
Sex: female (vs. male)	.431*** (.055)		.594*** (.081)		.727* (.111)	.726* (.110)
Another language spoken at home (vs. no)	1.107 (.243)		1.030 (.229)		1.137 (.279)	1.155 (.299)
Parent age at birth	.983 (.012)		.997 (.012)		.994 (.014)	.992 (.014)
Single-parent household (vs. two)	.878 (.144)		.775 (.142)		.738 (.141)	.767 (.151)
Number of siblings	1.283*** (.064)		1.114 (.064)		1.092 (.078)	1.116 (.079)
Parent long-standing MBN disorder (vs. no)	1.654* (.405)		1.601 (.416)		1.303 (.332)	1.357 (.367)
Birth weight	.643*** (.074)		.764* (.091)		.734* (.103)	.728* (.108)
Special care at birth (vs. no)	1.359 (.264)		1.268 (.252)		1.129 (.262)	1.065 (.265)
Aggregate health deprivation and disability decile	.997 (.026)		.991 (.026)		.986 (.029)	.992 (.029)
School special education need for SL (Age 7)						15.517*** (5.562)

Note: Exponentiated coefficients with standard errors in parentheses. Weighted for complex survey design and attrition. N = 6,498. LSL = linking sounds and letters; MBN = mental, behavioral, and neurodevelopmental; SL = speech-language.

*p < .05. **p < .01. ***p < .001.

As expected, the bivariate relationship between the FSP score and teacher recognition of speech-language needs at age 7 was significant in model 6. Each one-unit increase in the FSP score was associated with a 44.2 percent decrease in teacher odds of recognition (p < .001). Relative to model 5, the associations in model 7 with parental education, number of siblings, and parental MBN disorder were nonsignificant, and inconsistent income associations emerged. Sex and birth weight remained significant. The score itself remained similar in magnitude at a 41.5 percent reduction in odds of teacher recognition (p < .001).

Model 8 includes only the objective FSP LSL score and parental speech-language concern, both from age 5. Every one-unit increase in the FSP score was associated with a 40.3 percent decrease in the odds of teacher recognition (p < .001). Odds of teacher recognition of a child having speech-language needs was 10.0 times higher if the parent expressed speech-language concerns at age 5 (p < .001). The magnitude of the association is striking given that the teacher’s recognition of need, and thus the ability to act as an equalizer, should be based on objective health assessments. In model 9, we included all baseline background characteristics. Although we might expect the magnitude of the assessment score upon which teacher recognition of need should be based to stay similar, the magnitude of parental concern was also virtually unchanged with the inclusion of these additional predictors. The magnitude of sex was decreased by a considerable amount from model 7 to 9 (by 38 percent in the log-odds scale) with the inclusion of parental concern. Yet sex remained significant net of the objective assessment and parental concern, with teacher odds of identifying speech-language needs 27.3 percent lower for girls than boys (p < .05). Birth weight also remained significant. Although we caution that both teacher recognition and school SEN are measured at the same age and thus the direction cannot be established, model 10 shows that our results stand even controlling for having a SEN for speech-language.⁴

In Figure 2, we consider how the objective FSP speech-language assessment and parental concern worked together to influence teacher recognition of speech-language needs. The differences noted in the descriptive statistics in Table 2 were nearly reproduced, even net of background characteristics. The gap in the marginal predicted probability of teacher recognition was significant at all 10 score intervals. At the lowest score of 0, the predicted probability of teacher recognition was 0.79 if parents have speech-language concerns but only 0.29 if they did not. At scores 1, 2, and 3, indicating being behind learning goals, the respective probability of teacher recognition if parents were concerned was 0.70, 0.59, and 0.48 relative to 0.20, 0.14, and 0.09 if parents were not concerned. For a health-related issue for which assessment scores should drive service needs, these gaps are dramatic. Although the predicted probabilities were lower at the high end of the score distribution, the majority of students were located there (see Table 2), such that these probabilities amount to a large number of students who may be identified as having needs who likely do not simply because their parents expressed concern. And at this range of the distribution, such concern was tied to fundamental causes (Table 4, model 4). As an example, for a score of 7, where the most respondents were located (19.1 percent), the predicted probability of teacher recognition was 0.19 for students with parental concern and 0.02 for those without.

Figure 2.

Predicted probability of teacher recognition of child speech-language needs at age 7 by parental speech-language concern and foundation stage profile speech-language score at age 5 (Table 5, model 9).

We present three alternative specifications in Appendix A to consider the robustness of these findings. Panel A included an interaction between the FSP score and parental concern, which was nonsignificant (b = 0.099, p > .05). As would be likely then, the figure was virtually identical when an interaction is permitted. We note that because of the nonlinearity of the logit transformation, an interaction is not necessary for marginal predicted probabilities to differ in magnitude across two predictors, as was clearly the case in Figure 2, in which the association was much more consequential on the low end of the scale. In panel B, we also included an interaction, but used a categorical version of the FSP score. As can be seen comparing with Figure 2, the continuous version with no interaction captured the trend in this least parsimonious specification in panel B extremely well. Indeed, the lower Bayesian information criterion (BIC) and Akaike information criterion (AIC) for model 9 in Table 5 (BIC = 2,441.7, AIC = 2,299.36) indicated that the more parsimonious continuous model was a better fit than that from panel B (BIC = 2,566.4, AIC = 2,308.8). Finally, panel C shows a version of the figure that used the standardized version of the FSP LSL score, again leading to similar conclusions.

Logistic Regression of Age 11 School SEN for Speech-Language

In supplementary models, we also considered whether the child had a school SEN for speech-language needs at age 11 (sweep 5), shown in Appendix B. Although we caution that parental concern is separated by considerable time here, the models nonetheless demonstrate associations with later SEN status. The odds ratio for parental concern is reduced once teacher recognition is included in model B2 (from 13.1 to 4.9, p < .001). Yet parental concern remains significant, and this is the case even controlling for whether the child already had an SEN for speech-language age 7 in model B3 (odds ratio = 3.7, p < .05).

Discussion

Although “language” has received attention from sociologists when it represents a second language (e.g., Canizales 2021) or patterns of speech related to race or class (e.g., Carter 2006; Holland 2012; Lareau 2003; McGrady and Reynolds 2013; Morris 2005a, 2005b) that could result in discrimination in school settings, research on diagnosable and treatable speech-language disorders has remained understudied within sociology (see Hibel et al. 2010; Skrtic et al. 2021). Yet speech-language skills represent fundamental building blocks to all subject learning, such that those who do not acquire these skills experience poorer educational outcomes (Elbro et al. 2011; McGregor 2020). Unlike most other childhood health disorders, schools represent a primary site of recognition of speech-language issues as well as a site of intervention through speech-language pathologists. We bridge theories from sociology of education and medical sociology to examine the role of inequality in teacher recognition. With the exception of bias on the basis of student gender, we largely find that inequality in teacher recognition operates through parents. Parental concern about their children’s speech-language skills has an outsized association with whether teachers state the student has a speech-language issue. Students who have actual needs and are drawn disproportionately from disadvantaged backgrounds are less likely to be recognized without parental concern, whereas some students with no apparent need are being “recognized” because of likely unjustified concern of parents from advantaged backgrounds. Here, we consider the theoretical and practical implications of these findings.

In line with past research, speech-language assessment scores at age 5 were associated with several background socioeconomic and demographic characteristics. The findings for gender show that some teacher bias may remain regarding female students even after controlling for the assessment score, especially given our results show that females on average scored higher on the assessment. That is, teachers are less likely to recognize speech-language issues among female students. As communication expectations work to hide language difficulties in girls (Hart et al. 2004; Toseeb et al. 2017) and preschool girls are more often told to be quiet than boys because of gendered behavior expectations (Martin 1998), teachers may overlook girls’ speech-language needs. By contrast, although confirming prior research (Roy et al. 2014) that scores were lower for students from lower income, lower education, and geographically disadvantaged households, we largely found that teacher recognition was not associated with these measures of inequality once the assessment score was included in our model. This finding suggests that teachers are not using such class-based background characteristics as a direct factor in recognition of student needs. Rather, these forms of inequality appear to operate indirectly through parents.

Given the magnitude of the relationship, the role of parents in teacher recognition of speech-language issues was dramatic. Parental concern regarding speech-language skills at age 5 was associated with about 10 times higher odds of teacher recognition at age 7. Although omitted variable bias remains possible, the associations with parental concern were virtually unchanged when available background variables were included and even when controlling for school SEN, and we find it unlikely that there is a measure that would explain such a large association with parental concern. This relationship occurred throughout the FSP LSL score distribution. At the low end, where scores 3 and below indicate unsatisfactory age 5 speech-language progression, there were few distinguishing background characteristics, with two exceptions. First, gender once again was a source of inequality. Although all parents with children scoring 3 and below should be concerned, parents express less concern for their low-scoring female children. Although we cannot speak to the precise mechanism, parental gendered educational expectations regarding boys outperforming girls (e.g., Furnham, Reeves, and Budhani 2002; Kane 2012; Raley and Bianchi 2006; Tenenbaum and Leaper 2003), as well as gendered social activities that magnify boys’ yet hide girls’ speech-language difficulties (Hart et al. 2004; Toseeb et al. 2017), may result in reduced concern for girls and enhanced concern for boys among those behind.

Second, there was a large association with the parent having a long-standing MBN disorder (see Campbell et al. 2003), and this association emerged despite it not being predictive of the assessment score itself. Thus, in the part of the distribution where a child is likely in need of services, parents with experience with related disorders may be more able to advocate for their child. This is a novel finding with regard to fundamental causes: although fundamental causes theory usually speaks to sources of advantage, having an MBN disorder, which is often highly stigmatized, may help produce knowledge that allows such parents to convey their children’s needs to teachers. Although novel to medical sociology, this finding aligns with work at the intersection of sociology of education and sociology of race/ethnicity regarding how marginalized communities are able to build on “community cultural wealth” (Acevedo and Solorzano 2023; Yosso 2005). This concept challenges that marginalized communities lack cultural capital; rather, the cultural capital specific to marginalized communities that parents and students bring into the classroom can be used as a protective factor when encountering discrimination. According to Yosso (2005), this cultural wealth includes additional various forms of capital, including navigational, familial, resistant, and even linguistic capital. For parents with MBN disorders, they may draw on these sources to create the power and knowledge necessary to advocate for speech-language services because they can build on their own experience navigating and resisting discrimination in accessing services. That capital may also allow their children to advocate on their own behalf in the school setting. We encourage future research on this possible pathway.

The role of fundamental causes related to power, knowledge, and resources (Link and Phelan 1995) emerge most clearly at the high end of the FSP score distribution where students are highly unlikely to have a need for speech-language therapy. Here, White parents and parents with a bachelor’s degree express more concern for their high-scoring children.⁵ Why would parents whose children have seemingly no need to be recognized by teachers as having speech-language issues express concern? Lareau’s (2003) concerted cultivation offers two interdependent possibilities. First, given that such privileged parents try to concertedly cultivate their children through use of more demanding language patterns, they may misperceive their children’s inability to overperform as a speech-language issue that needs addressing. Second, advantaged parents feel a sense of entitlement to available services, thus pushing for services regardless of child need. In either scenario, teachers are reluctant to confront higher SES White parents and in the process may marginalize less affluent parents (Calarco 2020; Lewis and Diamond 2015; Posey-Maddox 2014; Wheeler 2018). As the home environment is stage setting for education (Harris and Robinson 2016), more advantaged children may also pick up on their parents’ concern and be more proactive in seeking help from teachers (Calarco 2011). Although we have no direct measure of the mechanism by which parental concern affects teacher recognition, the robust qualitative literature cited here demonstrates many of these very mechanisms in education, and we view our study as a quantitative companion. We encourage additional qualitative studies that specifically examine speech-language development and other areas where schools and health overlap. Finally, we reiterate the role that gender plays, as parents were more likely to express concern about high-scoring male children, which could potentially confer additional advantages onto overperforming boys (Musto 2019).

An open question then is whether schools are equalizing (Downey 2020) speech-language skills across sociodemographic categories. In a country such as England, where all students are screened for this health issue, it represents what medical sociologists call an “upstream intervention.” Such interventions are meant to apply equally across SES, thereby bypassing the power, knowledge, and resources related to fundamental causes of health disparities often required to take advantage of interventions. The education system is considered a primary upstream point for such health interventions (Hummer and Hernandez 2013; Zajacova and Lawrence 2018). One positive finding here is that teacher recognition appears driven largely by the objective assessment score and not directly by bias toward the student, although gender clearly remains an area in need of addressing. The negative finding is that teacher recognition is dramatically affected by parental concern, and parental concern is related to fundamental causes of health disparities. Thus, inequality in teacher recognition emerges through indirect pathways via parents, and supplemental models show that both have lasting associations with whether a student has an official school SEN for speech-language needs. As a result, some students who likely require speech-language services go unrecognized by their teachers as having such a need, which could further embed disadvantage given the known relationship between speech-language and academic outcomes (Elbro et al. 2011). Furthermore, students who are performing satisfactorily and even above expectations are recognized as having an issue simply because their advantageously positioned parents express concern. This parental action could be taking services away from those who need it, conferring additional advantages on the already advantaged, and creating an unnecessary staffing burden. At both ends of the spectrum, schools do not appear to act as equalizers across SES for speech-language issues because of the outsized role that parents play.

Limitations

We note limitations to our study. First, health needs may also be recognized outside the school setting within the medical system. However, there is little reason to assume parental concern would not operate there as well. The United Kingdom may also differ from the U.S. context given strong universal health care. We encourage similar studies in other national contexts.

Second, the FSP LSL is but one assessment of speech-language needs, and might be considered narrow relative to professional assessments by speech-language pathologists. Yet as the final of three universal assessments in England, it is still meant to flag children in need of further help, and any measurement error is highly unlikely to explain such a clear pattern and large association with parental concern on teacher recognition.

Third, although using measurements at ages 5 and 7 represents an advantage because parental concern cannot be affected by later teacher concern and the teacher conducting the FSP assessment at age 5 is extremely unlikely to be the age 7 teacher, a disadvantage could be that both parental concern could change and students could receive speech-language services that improve their skills between age 5 and 7. However, being behind at age 5 on speech-language skills is actually late for intervention, such that there is a moderate and stable relationship between age four-and-a-half speech-language scores and reading at 7, 9, and 10 years (Hayiou-Thomas et al. 2010). Furthermore, development between age 5 and 7 does not explain why for students who score high, parents are concerned or teachers are recognizing unjustified needs.

Finally, although careful to establish temporal order, we caution that we do not use causal methodology, such that all relationships should be interpreted as associations.

Conclusion

This study demonstrates that parents play an outsized role in whether teachers express that a child has speech-language issues, and that although social determinants of health affect speech-language assessment scores, such inequality mostly affects teacher recognition through parental concern. We note that our findings represent not only a call to sociologists to consider speech-language issues, but also a call to speech-language researchers to more directly consider the role of inequality. Only in 2022 did an article appear in one of the flagship journals within speech-language pathology calling upon researchers to consider the social determinants of language (Di Sante and Potvin 2022). Relative to sociology, as well as other health subfields, this call is surprisingly recent. Thus, both disciplines would benefit from interdisciplinary partnerships, such as that represented here. By building on joint expertise to address inequalities, we can uncover discrepancies in recognition of speech-language needs to provide children with the training they need to succeed both in early education and beyond.

Footnotes

Appendix

Appendix B.

Logistic Regression of School Special Education Need for SL at Age 11.

	Model B1	Model B2	Model B3
Foundation stage profile LSL score (age 5)	.620*** (.036)	.740*** (.042)	.776*** (.050)
Parental speech-language concern (age 5)	13.112*** (4.479)	4.924*** (2.262)	3.675* (1.909)
Country (vs. England)
Wales	1.964 (1.021)	1.858 (.912)	1.667 (.872)
Scotland	1.352 (.623)	1.158 (.557)	1.291 (.678)
Northern Ireland	.654 (.288)	.829 (.418)	.700 (.425)
Income quintile (vs. first)
Second	1.258 (.763)	1.510 (.940)	1.643 (1.120)
Third	1.022 (.792)	1.204 (.932)	.879 (.676)
Fourth	1.785 (1.202)	2.088 (1.444)	1.783 (1.345)
Fifth	2.138 (1.622)	2.320 (1.766)	2.002 (1.642)
Parent bachelor’s degree or higher	1.877 (.763)	2.630* (1.137)	3.184* (1.512)
Race: White (vs. non-White)	.956 (.565)	1.102 (.653)	1.439 (.765)
Sex: female (vs. male)	.900 (.294)	1.010 (.350)	1.102 (.391)
Another language spoken at home (vs. no)	.904 (.686)	1.102 (.810)	1.232 (.855)
Parent age at birth	1.073* (.032)	1.083** (.032)	1.074* (.033)
Single-parent household (vs. two)	.567 (.302)	.672 (.380)	.851 (.511)
Number of siblings	.866 (.129)	.842 (.155)	.840 (.156)
Parent long-standing MBN disorder (vs. no)	.880 (.502)	.760 (.438)	.762 (.527)
Birth weight	.537* (.159)	.621 (.203)	.523 (.176)
Special care at birth (vs. no)	.497 (.258)	.451 (.238)	.338 (.199)
Aggregate health deprivation and disability decile	.916 (.049)	.914 (.049)	.933 (.053)
Teacher recognition of child SL needs (age 7)		15.139*** (7.524)	9.207*** (4.850)
School special education need for SL (age 7)			10.158*** (5.390)

Note: Exponentiated coefficients with standard errors in parentheses. N = 5,791. LSL = linking sounds and letters; MBN = mental, behavioral, and neurodevelopmental; SL = speech-language.

p < .05. **p < .01. ***p < .001.

Acknowledgements

We wish to thank Doug Downey, Jacob Hibel, Lee Elliott Major, Rin Reczek, and Jeremy Staff for helpful comments.

ORCID iDs

Janet Vuolo

Mike Vuolo

Notes

Author Biographies

Janet Vuolo is an assistant professor in the Department of Speech and Hearing Science at The Ohio State University. Her research interests include speech and language development and disorders, motor speech disorders, and childhood apraxia of speech.

Mike Vuolo is a professor in the Department of Sociology at The Ohio State University and editor of Sociological Methodology, the American Sociological Association’s methods journal. His research interests include crime, law, and deviance; health; work; substance use; and methodology.

References

Acevedo

Nancy

Solorzano

Daniel G.

2023. “An Overview of Community Cultural Wealth: Toward a Protective Factor against Racism.” Urban Education 58(7):1470–88.

Al Otaiba

Stephanie

Puranik

Cynthia S.

Ziolkowski

Robyn A.

Montgomery

Tricia M.

2009. “Effectiveness of Early Phonological Awareness Interventions for Students with Speech or Language Impairments.” Journal of Special Education 43(2):107–28.

ASHA (American Speech-Language-Hearing Association). 1993. “Definitions of Communication Disorders and Variations.” Relevant Paper. Rockville, MD: American Speech-Language-Hearing Association.

ASHA (American Speech-Language-Hearing Association). 2003. “American English Dialects.” Technical Report. Rockville, MD: American Speech-Language-Hearing Association.

Buchmann

Claudia

Condron

Dennis J.

Roscigno

Vincent J.

2010. “Shadow Education, American Style: Test Preparation, the SAT and College Enrollment.” Social Forces 89(2):435–61.

Calarco

Jessica McCrory

. 2011. “‘I Need Help!’ Social Class and Children’s Help-Seeking in Elementary School.” American Sociological Review 76(6):862–82.

Calarco

Jessica McCrory

. 2020. “Avoiding Us versus Them: How Schools’ Dependence on Privileged ‘Helicopter’ Parents Influences Enforcement of Rules.” American Sociological Review 85(2):223–46.

Campbell

Thomas F.

Dollaghan

Christine A.

Rockette

Howard E.

Paradise

Jack L.

Feldman

Heidi M.

Shriberg

Lawrence D.

Sabo

Diane L.

Kurs-Lasky

Marcia

. 2003. “Risk Factors for Speech Delay of Unknown Origin in 3-Year-Old Children.” Child Development 74(2):346–57.

Canizales

Stephanie L.

2021. “Educational Meaning Making and Language Learning: Understanding the Educational Incorporation of Unaccompanied, Undocumented Latinx Youth Workers in the United States.” Sociology of Education 94(3):175–90.

10.

Carter

Prudence L.

2006. “Straddling Boundaries: Identity, Culture, and School.” Sociology of Education 79(4):304–28.

11.

Chang

Virginia W.

Lauderdale

Diane S.

2009. “Fundamental Cause Theory, Technological Innovation, and Health Disparities: The Case of Cholesterol in the Era of Statins.” Journal of Health and Social Behavior 50(3):245–60.

12.

Condron

Dennis J.

2007. “Stratification and Educational Sorting: Explaining Ascriptive Inequalities in Early Childhood Reading Group Placement.” Social Problems 54(1):139–60.

13.

Condron

Dennis J.

2009. “Social Class, School and Non-school Environments, and Black/White Inequalities in Children’s Learning.” American Sociological Review 74(5):685–708.

14.

Condron

Dennis J.

Downey

Douglas B.

Kuhfeld

Megan

. 2021. “Schools as Refractors: Comparing Summertime and School-Year Skill Inequality Trajectories.” Sociology of Education 94(4):316–40.

15.

Di Sante

Mélissa

Potvin

Louise

. 2022. “We Need to Talk about Social Inequalities in Language Development.” American Journal of Speech-Language Pathology 31(4):1894–97.

16.

Dowd

Jennifer Beam

Aiello

Allison E.

2008. “Did National Folic Acid Fortification Reduce Socioeconomic and Racial Disparities in Folate Status in the US?” International Journal of Epidemiology 37(5):1059–66.

17.

Downey

Douglas B.

2020. How Schools Really Matter. Chicago: University of Chicago Press.

18.

Downey

Douglas B.

Kuhfeld

Megan

van Hek

Margriet

. 2022. “Schools as a Relatively Standardizing Institution: The Case of Gender Gaps in Cognitive Skills.” Sociology of Education 95(2):89–109.

19.

Downey

Douglas B.

Pribesh

Shana

. 2004. “When Race Matters: Teachers’ Evaluations of Students’ Classroom Behavior.” Sociology of Education 77(4):267–82.

20.

Downey

Douglas B.

Quinn

David M.

Alcaraz

Melissa

. 2019. “The Distribution of School Quality: Do Schools Serving Mostly White and High-SES Children Produce the Most Learning?” Sociology of Education 92(4):386–403.

21.

Downey

Douglas B.

von Hippel

Paul T.

Broh

Beckett A.

2004. “Are Schools the Great Equalizer? Cognitive Inequality during the Summer Months and the School Year.” American Sociological Review 69(5):613–35.

22.

Elbro

Carsten

Dalby

Mogens

Maarbjerg

Stine

. 2011. “Language-Learning Impairments: A 30-Year Follow-Up of Language-Impaired Children with and without Psychiatric, Neurological and Cognitive Difficulties.” International Journal of Language & Communication Disorders 46(4):437–48.

23.

Forrest

Clair L.

Gibson

Jenny L.

Halligan

Sarah L.

St Clair

Michelle C.

2020. “A Cross-Lagged Analysis of Emotion Regulation, Peer Problems, and Emotional Problems in Children with and without Early Language Difficulties: Evidence from the Millennium Cohort Study.” Journal of Speech, Language, and Hearing Research 63(4):1227–39.

24.

Furnham

Adrian

Reeves

Emma

Budhani

Salima

. 2002. “Parents Think Their Sons Are Brighter Than Their Daughters: Sex Differences in Parental Self-Estimations and Estimations of Their Children’s Multiple Intelligences.” Journal of Genetic Psychology 163(1):24–39.

25.

Gansen

Heidi M.

2021. “Disciplining Difference(s): Reproducing Inequalities through Disciplinary Interactions in Preschool.” Social Problems 68(3):740–60.

26.

Glied

Sherry

Lleras-Muney

Adriana

. 2008. “Technological Innovation and Inequality in Health.” Demography 45(3):741–61.

27.

Guralnick

Michael J.

2011. “Why Early Intervention Works: A Systems Perspective.” Infants and Young Children 24(1):6–28.

28.

Hanselman

Paul

. 2018. “Do School Learning Opportunities Compound or Compensate for Background Inequalities? Evidence from the Case of Assignment to Effective Teachers.” Sociology of Education 91(2):132–58.

29.

Hanselman

Paul

Fiel

Jeremy E.

2017. “School Opportunity Hoarding? Racial Segregation and Access to High Growth Schools.” Social Forces 95(3):1077–1104.

30.

Harris

Angel L.

2011. Kids Don’t Want to Fail: Oppositional Culture and Black Students’ Academic Achievement. Cambridge, MA: Harvard University Press.

31.

Harris

Angel L.

Robinson

Keith

. 2016. “A New Framework for Understanding Parental Involvement: Setting the Stage for Academic Success.” Russell Sage Foundation Journal of the Social Sciences 2(5):186–201.

32.

Hart

Kerstine I.

Fujiki

Martin

Brinto

Bonnie

Hart

Craig H.

2004. “The Relationship between Social Behavior and Severity of Language Impairment.” Journal of Speech, Language, and Hearing Research 47(3):647–62.

33.

Harvey

Peter Francis

. 2023. “‘Everyone Thinks They’re Special’: How Schools Teach Children Their Social Station.” American Sociological Review 88(3):493–521.

34.

Hayiou-Thomas

Marianna E.

Harlaar

Nicole

Dale

Philip S.

Plomin

Robert

. 2010. “Preschool Speech, Language Skills, and Reading at 7, 9, and 10 Years: Etiology of the Relationship.” Journal of Speech, Language, and Hearing Research 53(2):311–32.

35.

Hernandez

Elaine M.

2013. “Provider and Patient Influences on the Formation of Socioeconomic Health Behavior Disparities among Pregnant Women.” Social Science & Medicine 82:35–42.

36.

Hernandez

Elaine M.

Vuolo

Mike

Frizzell

Laura C.

Kelly

Brian C.

2019. “Moving Upstream: The Effect of Tobacco Clean Air Restrictions on Educational Inequalities in Smoking among Young Adults.” Demography 56(5):1693–1721.

37.

Hibel

Jacob

Farkas

George

Morgan

Paul L.

2010. “Who Is Placed into Special Education?” Sociology of Education 83(4):312–32.

38.

Hoff

Erika

. 2003. “The Specificity of Environmental Influence: Socioeconomic Status Affects Early Vocabulary Development Via Maternal Speech.” Child Development 74(5):1368–78.

39.

Hogan

Tiffany P.

Catts

Hugh W.

Little

Todd D.

2005. “The Relationship between Phonological Awareness and Reading.” Language, Speech, and Hearing Services in Schools 36(4):285–93.

40.

Holland

Megan M.

2012. “Only Here for the Day: The Social Integration of Minority Students at a Majority White High School.” Sociology of Education 85(2):101–20.

41.

Hummer

Robert A.

Hernandez

Elaine M.

2013. “The Effect of Educational Attainment on Adult Mortality in the United States.” Population Bulletin 68(1):1–16.

42.

Ikeda

Joanne P.

Crawford

Patricia B.

Woodward-Lopez

Gail

. 2006. “BMI Screening in Schools: Helpful or Harmful.” Health Education Research 21(6):761–69.

43.

Irizarry

Yasmiyn

. 2015. “Selling Students Short: Racial Differences in Teachers’ Evaluations of High, Average, and Low Performing Students.” Social Science Research 52:522–38.

44.

Johnson

Odis

Wagner

Michael

. 2017. “Equalizers or Enablers of Inequality? A Counterfactual Analysis of Racial and Residential Test Score Gaps in Year-Round and Nine-Month Schools.” Annals of the American Academy of Political and Social Science 674(1):240–61.

45.

Kane

Emily W.

2012. The Gender Trap: Parents and the Pitfalls of Raising Boys and Girls. New York: New York University Press.

46.

Lareau

Annette

. 2003. Unequal Childhoods: Class, Race, and Family Life. Berkeley: University of California Press.

47.

Lewis

Amanda E.

Diamond

John B.

2015. Despite the Best Intentions: How Racial Inequality Thrives in Good Schools. New York: Oxford University Press.

48.

Link

Bruce G.

Phelan

Jo C.

1995. “Social Conditions as Fundamental Causes of Disease.” Journal of Health and Social Behavior 35(Extra Issue):80–94.

49.

Martin

Karin A.

1998. “Becoming a Gendered Body: Practices of Preschools.” American Sociological Review 63(4):494–511.

50.

Masters

Ryan K.

Link

Bruce G.

Phelan

Jo C.

2015. “Trends in Educational Gradients of ‘Preventable’ Mortality: A Test of Fundamental Cause Theory.” Social Science & Medicine 127:19–28.

51.

McGrady

Patrick B.

Reynolds

John R.

2013. “Racial Mismatch in the Classroom: Beyond Black-White Differences.” Sociology of Education 86(1):3–17.

52.

McGregor

Karla K.

2020. “How We Fail Children with Developmental Language Disorder.” Language, Speech, and Hearing Services in Schools 51(4):981–92.

53.

Meschi

Elena

Micklewright

John

Vignoles

Anna

Lindsay

Geoff

. 2010. “The Transitions between Categories of Special Educational Needs of Pupils with Speech, Language and Communication Needs (SLCN) and Autism Spectrum Disorder (ASD) as They Progress through the Education System.” Manchester: UK Department for Education.

54.

Morris

Edward W.

2005a. “From ‘Middle Class’ to ‘Trailer Trash’: Teachers’ Perceptions of White Students in a Predominately Minority School.” Sociology of Education 78(2):99–121.

55.

Morris

Edward W.

2005b. “‘Tuck in That Shirt!’ Race, Class, Gender, and Discipline in an Urban School.” Sociological Perspectives 48:25–48.

56.

Morris

Edward W.

2007. “‘Ladies’ or ‘Loudies’? Perceptions and Experiences of Black Girls in Classrooms.” Youth & Society 38:490–515.

57.

Musto

Michela

. 2019. “Brilliant or Bad: The Gendered Social Construction of Exceptionalism in Early Adolescence.” American Sociological Review 84(3):369–93.

58.

National Academies of Sciences, Engineering, and Medicine. 2016. Speech and Language Disorders in Children: Implications for the Social Security Administration’s Supplemental Security Income Program. Washington, DC: National Academies Press.

59.

Owens

Jayanti

. 2022. “Double Jeopardy: Teacher Biases, Racialized Organizations, and the Production of Racial/Ethnic Disparities in School Discipline.” American Sociological Review 87(6):1007–48.

60.

Polonijo

Andrea N.

2020. “The Impact of School-Entry Mandates on Social Inequalities in Human Papillomavirus Vaccination.” SSM – Population Health 12:100647.

61.

Polonijo

Andrea N.

Carpiano

Richard M.

2013. “Social Inequalities in Adolescent Human Papillomavirus (HPV) Vaccination: A Test of Fundamental Cause Theory.” Social Science & Medicine 82:115–25.

62.

Posey-Maddox

Linn

. 2014. When Middle-Class Parents Choose Urban Schools. Chicago: University of Chicago Press.

63.

Public Health England. 2020. “Best Start in Speech, Language and Communication.” London: Department of Health & Social Care.

64.

Raley

Sara

Bianchi

Suzanne

. 2006. “Sons, Daughters, and Family Processes: Does Gender of Children Matter?.” Annual Review of Sociology 32(1):401–421.

65.

Ready

Douglas D.

Chu

Elizabeth M.

2015. “Sociodemographic Inequality in Early Literacy Development: The Role of Teacher Perceptual Accuracy.” Early Education and Development 26(7):970–87.

66.

Roy

Penny

Chiat

Shula

Dodd

Barbara

. 2014. Language and Socioeconomic Disadvantage: From Research to Practice. London: City University London.

67.

Skrtic

Thomas M.

Saatcioglu

Argun

Nichols

Austin

. 2021. “Disability as Status Competition: The Role of Race in Classifying Children.” Socius. doi:10.1177/23780231211024398.

68.

St. Clair

Michelle C.

Forrest

Claire L.

Yew

Shaun Goh Kok

Gibson

Jenny L.

2019. “Early Risk Factors and Emotional Difficulties in Children at Risk of Developmental Language Disorder: A Population Cohort Study.” Journal of Speech, Language, and Hearing Research 62(8):2750–71.

69.

Tenenbaum

Harriet R.

Leaper

Campbell

. 2003. “Parent-Child Conversations about Science: The Socialization of Gender Inequities?” Developmental Psychology 39(1):34–47.

70.

Tomblin

J. Bruce

Records

Nancy L.

Buckwalter

Paula

Zhang

Xuyang

Smith

Elaine

O’Brien

Marlea

. 1997. “Prevalence of Specific Language Impairment in Kindergarten Children.” Journal of Speech, Language, and Hearing Research 40(6):1245–60.

71.

Toseeb

Umar

Pickles

Andrew

Durkin

Kevin

Botting

Nicola

Conti-Ramsden

Gina

. 2017. “Prosociality from Early Adolescence to Young Adulthood: A Longitudinal Study of Individuals with a History of Language Impairment.” Research in Developmental Disabilities 62:148–59.

72.

Tyson

Karolyn

. 2011. Integration Interrupted: Tracking, Black Students, and Acting White after Brown. New York: Oxford University Press.

73.

von Hippel

Paul T.

Workman

Joseph

Downey

Douglas B.

2018. “Inequality in Reading and Math Skills Forms Mainly before Kindergarten: A Replication, and Partial Correction, of ‘Are Schools the Great Equalizer?’” Sociology of Education 91(4):323–57.

74.

Wheeler

Sharon

. 2018. “‘Essential Assistance’ versus ‘Concerted Cultivation’: Theorising Class-Based Patterns of Parenting in Britain.” Pedagogy, Culture & Society 26(3):327–44.

75.

Yawn

Barbara P.

Yawn

Roy A.

Hodge

David

Kurland

Margary

Shaughnessy

William J.

Ilstrup

Duane

Jacobsen

Steven J.

1999. “A Population-Based Study of School Scoliosis Screening.” JAMA 282(15):1427–32.

76.

Yosso

Tara J.

2005. “Whose Culture Has Capital? A Critical Race Theory Discussion of Community Cultural Wealth.” Race, Ethnicity, and Education 8(1):69–91.

77.

Zajacova

Anna

Lawrence

Elizabeth M.

2018. “The Relationship between Education and Health: Reducing Disparities through a Contextual Approach.” Annual Review of Public Health 39:273–89.

78.

Zhang

Xuyang

Tomblin

J. Bruce

. 2000. “The Association of Intervention Receipt with Speech-Language Profiles and Social-Demographic Variables.” American Journal of Speech-Language Pathology 9(4):345–57.

How Social Determinants of Health Shape Parental Concern and Teacher Recognition of Student Speech-Language Needs

Abstract

Keywords

Social Determinants of Speech-Language Disorders and Schools as Equalizers

Schools and Teachers as Perpetuating Inequality

The Role of Parents and Inequality

Methods

Data

Variables

FSP Speech-Language Score

Age 5 and Age 3 Parental Speech-Language Concern

Age 7 Speech-Language SEN

Age 7 Teacher Recognition of Speech-Language Needs

Baseline Background Characteristics

Analysis

Results

Descriptive Statistics

Regression of FSP Speech-Language Score at Age 5

Logistic Regression of Parental Speech-Language Concern at Age 5

Logistic Regression of Teacher Recognition of Speech-Language Needs at Age 7

Logistic Regression of Age 11 School SEN for Speech-Language

Discussion

Limitations

Conclusion

Footnotes

Appendix

Acknowledgements

ORCID iDs

Notes

Author Biographies

References