In 1983, the American Speech-Language-Hearing Association (ASHA) published a position statement on the topic of social dialects. A major point of the publication was to formally recognize the difference between language variation that is caused by normal linguistic processes (i.e., dialects) and variation that is caused by an atypical or disordered language system (i.e., language impairment). Through the position statement, ASHA also formally rebuked the practice of diagnosing and treating any dialect of a language as an impairment. The data used to support this stance were sociolinguistic findings regarding the systematic and complex nature of all dialects, including those that are socially stigmatized (see dialect speakers).

ASHA's position statement remains relevant today. Research on children's acquisition and use of different dialects is still a relatively new area of scientific endeavor, but small strides have been made. For example, there now exist numerous articles describing different dialects of English, and current publications about childhood language development, assessment, and treatment now routinely include discussions of dialect diversity. The different types of testing biases that may surface when the language tester and testee come from different linguistic or cultural backgrounds have also been described (Fagundes et al., 1998; Wilson, Wilson, and Coleman, 2000). A few traditional speech and language tests have added alternative scoring procedures for some nonstandard dialects (Garn-Nunn and Perkins, 1999; Ryner, Kelly, and Krueger, 1999). Alternative or new dialect scoring methods also have been created for the analysis of children's conversational language samples (Nelson, 1991; Stockman, 1996; McGregor et al., 1997; Seymour, Bland-Stewart, and Green, 1998).

Most of the advances listed started with methods and materials that were designed for speakers of standard English. Two different types of changes were then made to these methods. One change involved broadening the range of language forms that are considered normal by including the contrastive patterns of different dialects. The other type of change was to restrict the analysis to only the noncontrastive patterns. Contrastive patterns are those that show variation in surface structure across different dialects of a language (Seymour, Bland-Stewart, and Green, 1998). In English, one contrastive pattern is the copula/auxiliary be. In standard dialects of English, overt marking of be is obligatory in utterances such as “you are walking.” In other dialects, such as African-American English (AAE) and Southern White English (SWE), overt marking of be in this context is optional. As a result, “you walking” and “you are walking” are both felicitous in these dialects. The use of contrast analysis when working with language sample data is an example of an alternative assessment method that treats the contrastive patterns of different dialects as normal (McGregor et al., 1997). With this method, the only language patterns that can be viewed as errors are those that cannot occur in the child's dialect.

Noncontrastive patterns are those that do not show surface variation across different dialects of a language (Seymour, Bland-Stewart, and Green, 1998). One noncontrastive pattern of English is S-V-O word order (Martin and Wolfram, 1998). This pattern is thought to be noncontrastive because all dialects of English thus far have been shown to present this word order. Other patterns thought to be noncontrastive in English include various forms of complex syntax that make sentential coordination and subordination possible. An example of a language assessment method that restricts the analysis to the noncontrastive patterns of dialects is Stockman's (1996) Minimal Competency Core (MCC) analysis. The goal of MCC analysis is to identify and then evaluate a common core of language that can be found in multiple dialects of a language. As a criterion-referenced procedure, MCC specifies a minimum level of competency for each language pattern and each age level examined. One of the language items included in MCC is a child's mean length of utterance (MLU). For English-speaking 3-year-olds, Stockman (1996) sets the minimum MLU at 3.27 morphemes. She also lists 15 consonants in the initial position and a set of semantic expressions and pragmatic functions that all 3-year-olds should be able to demonstrate, regardless of the English dialect they use. Another example of a relatively new assessment tool that targets the noncontrastive features of dialects is that formulated by Craig, Washington, and Thompson-Porter (1998b), which uses Wh-questions and passive probes.

All of these advances treat the contrastive patterns of dialects as problematic for diagnostic purposes. The problem, as articulated by Seymour, Bland-Stewart, and Green (1998), is that some contrastive patterns of some nonstandard English dialects can look very similar to those that are produced by standard English-speaking children who have a language impairment. Some of the surface patterns that are generated by both language learning conditions include zero marking of be (e.g., “you walking”), zero marking of past tense (e.g., “yesterday she fall”), and zero marking of third person (e.g., “today he walk”). Seymour, Bland-Stewart, and Green refer to these patterns and other contrastive forms as presenting a diagnostic conundrum because interpretations of their use as markers of either a normal dialect or a grammatical impairment are difficult. These authors also state that the exclusion of the contrastive patterns within assessment is necessary only until more is known about children's acquisition and use of these patterns. As more research is completed, new methods that include the contrastive patterns should be made possible.

Recently, research has begun to focus on children's acquisition and use of the contrastive patterns of dialects. Findings from some of these studies suggest that these particular language forms may not be as problematic as they first seemed. For example, at least four studies have examined the effect of these patterns on standard calculations of children's average utterance length and utterance complexity. Each of these studies has shown children's use of the contrastive patterns to play a minimal role within these calculations. For three of the studies, the focus has been on the contrastive patterns of AAE (Craig, Washington, and Thompson-Porter, 1998a; Jackson and Roberts, 2001; Smith, Lee, and McDade, 2001). The participants in these studies have ranged in age from 3 to 9 years. Measures of length have been calculated on utterances, C-units, and T-units, and measures of complexity have involved counts of complex syntax. In every case, children's use of the contrastive patterns of AAE has been shown to be relatively unrelated to the length and complexity indices. For example, Jackson and Roberts (2001) report correlations between children's use of contrastive AAE patterns and their utterance length and complexity scores to be at or below –.11.

Oetting, Cantrell, and Horohov (1999) also examined the effect of contrastive dialect forms on standard calculations of utterance length and utterance complexity. The participants in their study were children who spoke a rural Louisiana version of SWE, and they ranged in age from 4 to 6 years. Approximately one-third of the children were classified as specifically language impaired (SLI); the others were classified as normal. Three language indices, MLU, developmental sentence score (DSS), and Index of Productive Syntax (IPSyn), were evaluated. To examine the effect of the contrastive forms, scores for MLU, DSS, and IPSyn were calculated twice for each child, once using samples that contained utterances with contrastive forms and once using the same samples with the contrastive utterances removed. Results indicated that the diagnostic classification of each child as either normal or SLI was the same, regardless of whether utterances with the contrastive forms were included or excluded.

Recent studies also suggest that grammatical weaknesses of children with SLI can be identified within the contrastive forms. In addition to evaluating indices of utterance length and complexity, Oetting, Cantrell, and Horohov (1999) examined the grammatical profile of SLI within the context of nine contrastive forms of SWE. For five of the forms (i.e., third person regular, contractible copula, contractible auxiliary, uncontractible auxiliary, and auxiliary does), the children with SLI were found to present rates of overt marking that were lower than those of their SWE-speaking age-matched and language-matched peers. In a second study, Oetting and McDonald (2001) examined the grammatical weaknesses of SLI in the context of two nonstandard dialects, SWE and a rural Louisiana version of AAE. In this study, 35 different contrastive patterns were coded. Differences between the normal children and those with SLI were identified for 14 of the contrastive patterns. A full model discriminant function that involved counts of all 35 patterns resulted in 90% of the children being correctly classified as either normal or impaired. Stepwise analyses yielded slightly different discriminant functions for identifying children with SLI in the SWE dialect as compared to the AAE dialect, but both models included language forms needed to formulate questions and mark tense. The finding that both dialect groups with SLI were shown to have trouble with these two areas of grammar is consistent with SLI studies that have been completed with standard English speakers, nonstandard English speakers, and speakers of languages other than English (e.g., Rice, Wexler, and Hershberger, 1998; Seymour, Bland-Stewart, and Green, 1998; Craig and Washington, 2000; Paradis and Crago, 2000).

A few recent studies also have examined the developmental tragectories of particular contrastive patterns in isolation (Wyatt, 1996; Henry et al., 1997; Jackson, 1998; Burns et al., 1999; Wynn, Eyles, and Oetting, 2000; Ross, Oetting, and Stapleton, in press). The contrastive patterns examined in these studies have included aspectual be and preterite had + Ved in AAE, copula be in AAE and SWE, and negative concord in Bristol English and Belfast English. Each of these studies has shown normally developing children to be remarkably capable of learning the distributional properties of their native dialect. This finding occurs even when children who speak different dialects of a language live in the same community and attend the same schools (Wynn, Eyles, and Oetting, 2000; Ross, Oetting, and Stapleton, in press). Of the studies listed that also included children with SLI, some group differences (normal versus impaired) have been identified, but the nature of these differences warrants further study.

Understanding the ways in which a childhood language impairment manifests within the contrastive and noncontrastive forms of different dialects is a topic of ongoing study. Additional work also is needed to extend the study of childhood language impairment to other language-learning situations. Two such situations are bilingual language acquisition and second language learning. Until this research is completed, our tools for identifying children with language weaknesses and our understanding of language impairment as a construct will remain limited.