| |
More than one in five children in the United States live in poverty, with pervasive consequences for their health and development. These consequences include effects on language. Children who live in poverty develop language at a slower pace than more advantaged children, and, after the age at which all children can be said to have acquired language, they differ from children from higher income backgrounds in their language skills and manner of language use. Children from low-income families are also overrepresented among children diagnosed as language-impaired.
These relations between poverty and language are robust, but they are complicated to describe and difficult to interpret. Poverty itself is not a homogeneous condition but occurs on a gradient and is usually associated with other variables that affect language, particularly education and ethnicity. Thus, the source of poverty effects is sometimes obscure. Further, the language children display is always a combined function of their language skill and language style. Thus, the nature of poverty effects is sometimes unclear. This article describes the observed associations between poverty and language development, and then considers why those associations occur. Much of the relevant literature does not use poverty per se as a variable but instead uses correlated variables such as education or a composite measure of socioeconomic status (SES). Thus, an important task in studying poverty effects is to identify the causal factors at work when poverty, low levels of parental education, and other associated factors are correlated with low levels of language achievement in children.
Language Differences Related to Income or SES
Before speech begins, infants who live in poverty produce speech sounds and babble in much the same way and on the same developmental timetable as all normally developing infants (Oller et al., 1995), and even at 3 years of age, SES is unrelated to the accuracy of children's articulation (Dollaghan et al., 1999). In contrast, virtually every other measure of language development reveals differences between children from low-income or low-SES families and children from higher income or higher SES families.
The clearest and largest SES-related difference is in the area of vocabulary. Recordings of spontaneous speech, maternal report measures, and standardized tests all show that children from low-income and low-SES families have smaller vocabularies than children from higher income and higher SES families (Rescorla, 1989; Hart and Risley, 1995; Dollaghan et al., 1999). By the age of 3 years, children from low-income families have productive vocabularies averaging around 500 words, while children from higher income families have productive vocabularies averaging more than 1000 words (Hart and Risley, 1995). Eighty percent of toddlers from low-income families score below the 50th percentile on the MacArthur Communicative Development Inventory (CDI) (Arriaga et al., 1998). Some findings suggest that low family income is more associated with measures of children's productive vocabulary than with measures of their receptive vocabulary (Snow, 1999). Two studies have reported SES-related differences in children's vocabulary, with lower SES children showing larger vocabularies. Both used the CDI, and both attributed their findings to lower SES mothers' tendencies to overestimate their children's abilities (Fenson et al., 1994; Feldman et al., 2000).
Grammatical development is also related to income or SES. Compared with higher SES children, children from lower social strata produce shorter responses to adult speech (McCarthy, 1930), score lower on standardized tests that include measures of grammatical development (Morrisset et al., 1990; Dollaghan et al., 1999), produce less complex utterances in spontaneous speech as toddlers (Arriaga et al., 1998) and at age 5 (Snow, 1999), and differ significantly on measures of productive and receptive syntax at age 6 (Huttenlocher et al., 2002). As an indicator of the magnitude of these effects on grammatical development, the low-income sample studied by Snow (1999) had an average MLU at age 3 years 9 months that would be typical of children more than a year younger, according to norms based on a middle-class sample. At age 5 years 6 months they had an average MLU typical of middle-class children age 3 years 1 month. On the other hand, the SES-related differences in productive syntax are not in whether children can or cannot use complex structures in their speech, but in the frequency with which they do so (Tough, 1982; Huttenlocher et al., in press). Studies of school-age children find SES-related differences in the communicative purposes to which language is put, such that children with less educated parents less frequently use language to analyze and reflect, to reason and justify, or to predict and consider alternative possibilities than children with more educated parents. The structural differences in children's language associated with SES may be a by-product of these functional differences (Tough, 1982).
SES-related differences in school-age children also appear in the ability to communicate meaning through language and to draw meaning from language, sometimes referred to as speaker and listener skills (Lloyd, Mann, and Peers, 1998). In the referential communication task, which requires children to describe one item in an array of objects so that a visually separated listener with the same array can identify that item, lower SES children are less able than higher SES children to produce sufficiently informative messages and to use information in messages addressed to them to make correct choices (Lloyd et al., 1998). Children from lower socioeconomic strata also perform less well than higher SES children in solving mathematics word problems. This poorer performance reflects a difference in language ability, not mathematical ability, because the same children show no difference in performance in math calculations (Jordan, Huttenlocher, and Levine, 1992).
The foregoing effects are effects of poverty or SES on variation within the normal range. On average, children from low-income families acquire language at a slower rate and demonstrate both differences in language use and poorer language skills than children from higher income families. Low SES is also a correlate of the diagnosis of specific language impairment (SLI) (Tomblin et al., 1997), although it is not clear what this means, given that SLI is defined in terms of delay relative to norms and that normative development is slower for lower SES children (Fazio, Naremore, and Connell, 1996). For this reason, there have been calls for sensitivity to SES effects on standardized measures in diagnosing SLI (Arriaga et al., 1998; Feldman et al., 2000), and efforts are underway to develop tests of language impairment that are direct tests of language learning ability rather than norm-referenced comparisons (Fazio, Naremore, and Connell, 1996; Campbell et al., 1997; Seymour, 2000).
Understanding the Relation Between Poverty and Language Development
Low family income cannot directly cause the depressed language skills associated with poverty but must operate via mediators that affect language. Potential mediators, or pathways, through which poverty operates may include factors with general effects on health and development, such as nutrition, exposure to environmental hazards, and quality of schools and child care, and may also include factors with specific effects on language, such as the opportunity for one-to-one contact with an adult (McCartney, 1984) and the language use of parents and classroom teachers (Huttenlocher et al., 2002).
A variety of evidence suggests that an important mediator of the relation between SES and language development in children is the nature of the language-learning environment provided by the family (Hoff, 2003) and that low levels of parental education, more than low income per se, affect the language-learning environment in the home. Two predictors of language development, the talk parents address to children and the exposure to books that parents provide, differ as a function of parental education. Compared with children of more educated parents, children of less educated parents hear less speech, hear less richly informative speech, receive less support for their own participation in conversation, and are read to less (Hart and Risley, 1995; Hoff-Ginsberg, 1998; U.S. Department of Education, 1998; Hoff, Laursen, and Tardif, 2002). When mediators of the relation between family SES and child language have been examined, properties of children's language-learning environments have been found to account for most of the variance attributable to SES both for syntax and for lexical development (Hoff, 2003; Huttenlocher et al., 2002). In fact, even variation within low-income samples is attributable to variation in language-learning environments. Among a group of 5-year-olds in Head Start programs (thus, children from low-income families), standard scores on a measure of comprehension vocabulary (the Peabody Picture Vocabulary Test) were significantly related to maternal use of sophisticated vocabulary (i.e., low-frequency words) and to the frequency of supportive mother-child interactions (Weizman and Snow, 2001). In a different sample of 4-year-olds in Head Start, variation in productive and comprehension vocabulary was accounted for by children's literacy experiences at home (Payne, Whitehurst, and Angell, 1994).
With respect to understanding the role of poverty in communication disorders, the literature presents a paradox. Poverty is associated with slow language development because poverty is associated with less supportive language-learning environments for children than more affluent situations. Poverty is also associated with the diagnosis of SLI, although most evidence suggests that the language environment is not the cause of SLI (Lederberg, 1980). In fact, studies of the heritability of language find a higher heritability for language impairment than for variation in language development within the normal range (Eley et al., 1999; Dale et al., 2000). If input is the mediator of poverty effects on language but input does not explain SLI, then why is poverty associated with SLI? The inescapable conclusion is that children can differ from the normative rate of development for two reasons: an impairment in the ability to learn language or an inadequate language-learning environment. Low percentile scores on measures of language development do not, by themselves, distinguish between these. While poverty does not cause language impairment or a communication disorder, the impoverished language-learning environment often associated with poverty can similarly impede language development.
See also specific language impairment in children.
| |
