The final panel session explored how language differences have been construed from psycholinguistic and sociolinguistic perspectives and the implications for understanding achievement gaps and designing instruction. Presenters pointed to a range of cognitive and social mechanisms through which speaking nonstandard varieties of English might affect achievement gaps, and considered the study of language differences, schooling, and achievement gaps from a historical perspective: What historical lessons might be applied to framing more useful discussions about research and practice for the future?
DIALECTS AND NONSTANDARD ENGLISH
William Labov and Anne Charity Hudley explored differences in language and achievement associated with language dialect (or vernacular). Labov began by summarizing the first part of the paper (Labov and Hudley, 2009), which focused on two main mechanisms by which linguistic factors associated with dialect may affect students’ academic achievement: (1) structural differences, phonemic inventory and grammatical rules that may interfere with reading and learning in standard English; and (2) symbolic influences, the social and psychological effects that result from the perceptions of teachers and others about the abilities and conduct of students who speak certain dialects. Labov and Hudley (2009) elaborate that, according to the 2009 National Assessment of Educational Progress (NAEP) report, both of these mechanisms may contribute to stable and
persistent minority gaps in literacy achievement and efforts to close these gaps may take different forms depending on which type of effect is most active.
According to Labov and Hudley (2009), the position held by most linguists and anthropologists is that all dialects learned by children as their first language have equal capacity for logical expression; the errors in reading and writing that those children make occur because of lack of alignment between a perfectly acquired vernacular and imperfectly acquired knowledge of the standard language of the classroom. A review of studies on African American vernacular English conducted from 1966 to 2002 showed considerable variation in pronunciation, while also showing what Labov described as “astonishing uniformity” in the grammatical structures of the dialect across the country.
The California Board of Education recently sponsored a consensus panel to determine the structural domains that differ most between standard English and the African American dialect with the goal of informing publishers about how aspects of African American vernacular English affect reading achievement (summarized in Labov and Hudley, 2009). Labov focused on two of the structural features that differentiate non-standard dialects from standard English to show how they appear to affect learning. These features are both grammatical suffixes among those present in standard English but absent in the African American vernacular: possessive –s added to the first of two nouns or noun phrases (as in “John’s house”) and the verbal –s attached to the third singular form of the verb in present tense (as in “He walks”). Other forms are more variably absent, such as usage of –ed to mark past tense (as in “He walked”) and copula apostrophes (as in the contraction, “He’s here.”). Possessive –s and other forms are not necessarily completely absent from the vernacular grammar, but they are used rarely or differently: for instance, possessive –s is found regularly when no other noun phrase follows: This is John’s; this is hers; this is mines. (Much more variability is found in the grammatical structures of Hispanic speakers across the country, Labov noted.)
How might the absence of these two features as used in standard English affect reading achievement? Would readers decode the form correctly when reading or would they not process the features as intended thereby affecting comprehension of the text?
Part of a longitudinal study at the Frank Porter Graham Child Development Center at the University of North Carolina used the Woodcock-Johnson Applied Problems Test to examine the relation between the linguistic complexity of word problems in math and correct computation
(see Terry et al., in press). Although many linguistic features—such as past –ed, auxiliary “have,” and possessive pronouns—did not relate to math computation, the possessive –s and the verbal –s correlated significantly math scores (r = .35 and .56, respectively). The mechanism behind the effect is not clear and it was concentrated in a small number of students for whom the forms were most often absent. In Labov’s view, these results show the importance of discovering which grammatical differences have cognitive effects and which do not so that effective teaching strategies can be developed, preferably when children are first learning to read.
From the earliest stages of inquiry into the effect of dialect differences, Labov said, there has been general agreement that in reading instruction, it is essential to distinguish mistakes in reading that affect comprehension, true reading errors, from differences in pronunciation that do not affect comprehension, such as reading “pen” as “pin.” The latter mistakes tend to be common in African American vernacular and other Southern dialects. An instructor who is aware of dialect differences would recognize this type of oral mistake as only a potential reading error. In another example, in diagnostic reading, the word “sneaked” is frequently read as “snuck” by speakers of many backgrounds. It is clear these readers have correctly decoded “sneaked” in order to produce “snuck.” Some errors are more difficult to distinguish, especially from pronunciation errors. If a student reads He was cold without the final /d/, He was [kol], there is no immediate way of knowing that the student did not mistake cold for coal, affecting the student’s ability to comprehend the text that follows.
As explained in Labov and Hudley (2009), a large part of instruction in the early acquisition of literacy comes through monitoring of oral reading. One task of a teacher is to recognize when the reader needs help decoding a word. When the teacher recognizes an error, the teacher may choose to intervene, supply the correct form, prompt for another reading, and perhaps follow with an explanation of the general principle involved. However, the problem of how to recognize a true reading error has been debated for some time. It is often assumed that a true reading error will lead to additional errors in the text that follows. If pen is pronounced as “pin” or sneak is read as snuck, it is unlikely to affect reading errors in the text that follows; in contrast, a truly misread word will affect the frequency of “following errors,” providing a way to estimate the number of true reading mistakes. If the absence of certain grammatical forms in a dialect leads children to decode words inaccurately, they may not ascertain the meaning conveyed in those forms and so the text will be difficult to comprehend.
Labov and Baker (in press) examine the probability of following errors for 155 African American and 186 Latino struggling student readers for omission of three kinds of grammatical suffixes: past –ed,
possessive –s, verbal –s and copula –s. When African American vernacular speakers made a clear error involving the absence of elements, such as verbal –s and possessive –s, for instance, rather than only potential errors (for instance, sneak versus snuck), the rate of errors when reading the text that follows just about doubles. The frequency of following errors for past –ed was not significantly different from correct readings, but for the other three error types there was a significant probability that the relation of that word to the rest of the sentence had not been ascertained. (Labov noted that Hispanic readers show a somewhat different pattern of errors.) Labov argued that these examples illustrate the importance of intervening “vigorously” to help students understand and correct grammatical inflections that affect comprehending text. Two research goals, he said, would be to learn more about how decoding grammatical signals affect sentence comprehension and to find new and better ways of teaching abstract features of standard English.
Symbolic (Social and Psychological) Influences
With regard to symbolic influences, Anne Charity Hudley listed several areas of linguistic variation observed in African American vernacular—differences in phonology and grammar, differences in vocabulary, differences in discourse and cultural patterns, and differences in self-presentation through language. As elaborated in Labov and Hudley (2009), these sources of significant structural and symbolic mismatches between classroom English and students’ language may influence: student confidence with reading and other attitudes toward reading and schooling; performance on standard assessments; teacher perceptions of student abilities, behavior, or both; and miscommunications between teachers and students. In turn, these effects can lead to expulsion, lower rates of achievement, and, ultimately, lower occupational attainment.
To remedy the effects of dialect on achievement, Hudley proposed greater sharing of information between academic researchers and K-12 educators so that educators can learn to correctly identify systemic patterns of home languages and dialects in their students, distinguish between speech disorders and nonstandard dialects, and make the educational system more accessible by teaching conventions while acknowledging the legitimacy of home languages and dialects. In this effort, a broad network of educators would need to be reached, especially in community colleges, historically black colleges and universities, and institutions that serve primarily Spanish-speaking students to communicate what is known about linguistics in an accessible manner to educators, including administrators, reading specialists, curriculum and instruction developers and supervisors, and speech pathologists.
Valdés and colleagues offered a historical perspective on how language has been studied in different disciplines, and emphasized three main points elaborated in Valdés, MacSwan, and Alvarez (2009). First, differences between cognitive and social interactionist views of language have implications for designing second-language instruction. Cognitively oriented theorists see language development as a change in mental state in which the knowledge of grammatical rules develops in a mostly linear fashion, and children come to use that knowledge with increasing complexity and control. Cognitively oriented researchers study differences in children’s usage and understanding of linguistic structure at different stages of language development, and how children cognitively process language, with a focus on the kinds of grammatical errors made at particular points in language acquisition.
In contrast, socially oriented researchers do not focus exclusively on the linguistic aspects of learning a second language, but strive for a broader understanding of how speakers of one language become users (speakers, writers, readers) of the second language. Of primary interest is how interactions between second-language learners and speakers of the second language can gradually support second-language learners in using an oral- and written-language system to communicate. Socially oriented researchers have contended, moreover, that much research from a cognitive perspective assumes a deficit perspective that emphasizes “learners’ limitations and their failure to become identical to native speakers (Valdés, MacSwan, and Alvarez, 2009). A deficit theory posits that a student who fails in school does so because of assumed internal deficits or deficiencies associated with limited intellectual abilities, linguistic shortcomings, lack of motivation to learn, and “immoral” behavior (Valencia, 1997).
Researchers from both perspectives agree, however, that when a particular language has been acquired the speakers will possess both linguistic knowledge (knowledge of the sound system, meanings of words, and syntactic rules) and pragmatic knowledge (how to participate in conversations and what to say to whom and when). They also agree that, whatever the mechanisms of learning are, they are shared across all humans, and how any particular language develops depends on the person’s specific language environment. The main disagreement lies in whether learning grammatical rules is the most essential aspect of learning language, and about whether those rules should be taught explicitly or set as a priority for assessment and instruction (for more background, see Valdés, MacSwan, and Alvarez, 2009).
A second major point was that several dichotomies have dominated and polarized the field of second-language acquisition and have led to labeling certain types of language as “good” or “bad” (see Valdés, MacSwan, and
Alvarez, 2009, for this analysis). Dichotomies pertaining to monolingual speakers include standard versus nonstandard English; standard English versus dialects; and elaborated code (complex, formal language used in academic conversation) versus restricted code (simple, fragmented language used in everyday conversation). Dichotomous perspectives on the language of bilingual speakers include additive bilingualism (proficiency in two languages, with positive cognitive benefits of knowing two languages) versus semi-lingualism (lack of competence in all languages an individual knows); context-reduced communication (does not assume shared experience, thus requiring precise elaboration) versus context-embedded communication (assumes shared experience, reducing the need for linguistic elaboration of the message); basic interpersonal communication skills versus cognitive academic language proficiency; and native speakers versus nonnative speakers (the native-nonnative distinction is imprecise and difficult to define in practice) (see Valdés, MacSwan, and Alvarez, 2009, for further analysis).
The most recent dichotomy, Valdés argued, contrasts academic language for school with ordinary, everyday language. To benefit students’ learning and to avoid past controversies associated with labeling particular kinds of language as either good or bad, the study of academic language needs to focus on documenting which aspects of language are truly important for learning academic content. The approach to studying academic language described by Schleppegrell (2009) is a promising way forward, Valdés said, though the definition of academic language remains much too broad and needs to be specified further through research. Until then, the term could be misused to refer simply to features of language that children with achievement gaps do not use, or, conversely, to features they possess but that “language-majority” children do not.
In future work on academic language, both researchers and educators need to recognize that academic language is contextualized: its development depends on understanding the context of language use and on the match between that context and the language students bring to school. In addition, most current work on academic language focuses on oral language rather than written language; a more balanced emphasis seems warranted.
Valdés concluded with a framework for having more productive discussions within and across the fields that study language in the context of schooling. First, researchers need to figure out how to “curricularize language,” which will be critically important for advancing instruction (see the discussion in Chapter 4). A barrier to developing and studying curriculum approaches, however, is that researchers of first and second-language acquisition and researchers of language pedagogy tend not to converse. These fields will need to work together more closely to identify
which language differences are important for school learning and also sensitive to instructional intervention, and to determine what the interventions should be.
Language research suggests that not all differences will be easy or possible to change or to teach regardless of how “teach” gets defined (whether more implicit or more explicit). It is also not yet clear whether the features of language that warrant targeting can be learned in classroom settings. To some extent the instructional approach desired may depend on what one counts as success. Explicit grammar instruction might help performance on standard grammatical tests, but a broader orientation to instruction would be needed to develop learners’ ability to use grammatical features in communication. Valdés also suggested rethinking whether monolingualism for students is a useful ideal to aspire to for the future. Finally, though language is vital to schooling, language is just one variable to consider in accounting for the academic disengagement and academic failure of many language-minority and disadvantaged students in the United States.
African American English
Lisa Green stressed the need in future research to become much more concrete in discussing notions of dialect and to move beyond the tendency of researchers to study only a sliver of language at one point in time. As Erika Hoff had mentioned earlier, ascertaining what speakers of various dialects know about language is currently more difficult than determining what speakers of only standard English know. Yet this information is needed to guide instruction, she said—and it is especially urgent to have for children starting at ages 4 or 5, when they first begin school—and to support early intervention.
Focusing on her own research relating to the African American dialect as an example, Green said a more thorough modeling of the structural features of African American English is needed to articulate how the features develop over time and how these developmental progressions relate to those found in “mainstream” English. This type of comprehensive, fine-grained, and developmental analysis of children’s speech can produce more accurate conceptualizations of African American English for targeting instructional practices that support language for achievement. At present, there have been few studies on the development of African American English (see Figure 5-1).
Green has examined children’s story narratives to document grammatical markers used by African American English speakers, including
the grammatical patterns unique to the dialect (e.g., habitual be as in “He be wearing a boot,” which means “He generally wears a boot”) and markers that are shared with the standard variety of English. This research is showing that forms of the dialect emerge in predictable developmental patterns, as is true of children who use only mainstream English. Exactly how these emerging African American dialect patterns relate to performance on mainstream language assessments is not yet clear, but the narratives are one way, Green suggested, to gain insight into what African American dialect speakers know about how to represent meanings and which meanings are represented at different points in their language development.
Results from Green’s research indicate that children acquire the language variations exhibited in their communities: both the dialect and mainstream English forms are acquired to different degrees depending upon their exposure to the language patterns of different communities, and the features of the dialect overlap with mainstream English. In addition, though it can appear that children do not know certain grammatical forms, such as those for representing past tense (–ed), a closer look reveals that the forms can appear in the African American dialect speech
but that they are used differently, with some of the forms even appearing much earlier in development than observed in children who speak only standard English. Children also may use both standard English and the dialect to mark meaning, such as past tense, in the same narrative. This knowledge is evident, she said, when children engage in variable shifting (using the same grammatical form available in one language for a different purpose in the second language) and code shifting (using a grammatical form not shared between two languages) to describe this movement between standard and dialect English.
Green argued that the appearance of variable shifting and code shifting in children’s language supports the notion that regularized African American English-language patterns are part of a linguistic system that is on a continuum with standard English patterns: they are not two dichotomous language varieties. Learning more about this continuum and the grammatical patterns and usage in the dialect in future research would help to determine whether variable shifting or code shifting is needed to increase usage of forms used most often in academic settings. Instruction may need to focus more on variable shifting than code shifting to encourage student awareness of when and where to use the forms they already possess. Some of the standard English forms that dialect speakers may need to learn for school might lie at the periphery on the continuum of grammar usages; if so, studies may show that it is beneficial for teachers to focus on developing those forms first since they may be more difficult to acquire.
In closing, Green suggested a model, referred to as the D.I.R.E.C.T. model (Green, in press) to guide interventions with teachers that would heighten their awareness about the development of language varieties, including African American English. The model specifies that teachers would benefit from knowing that the African American dialect is an inherently variable variety that has set patterns of use of grammatical markers. It is a native variety for many children, with regularized patterns of language use experienced in their environments. It is not haphazard language use or misuse of mainstream American English, Green stressed. African American and mainstream English express the same concepts by using different strategies of marking information about events. Educators can help students develop the standard variety of English for school by being aware of specific dialect markers and their meanings.
Latino Populations and Other Groups
Robert Bayley agreed that Labov’s more than 40 years of research on African American English and the research reviewed in Labov and Hudley (2009) effectively make the case for developing reading interven-
tions to assist struggling African American readers, and Hudley’s recent contributions on symbolic, social, and psychological influences extend that work in valuable ways for theory and practice. The relationship between dialect and decoding is important, and Labov’s work has shown that intervention can improve reading scores for African American dialect speakers.
Variations in Latino languages have not received much attention, however, in the work of Labov and others, Bayley observed. He suggested that more needs to be known about how the different backgrounds of Latino children have affected the reading error patterns that Labov reported for Latino students in comparison with African American students, the range of Latino grammatical structures, and the types of instruction that would be most effective and appropriate to support reading. For instance, were the Latino students bilingual to some degree? Spanish dominant? Literate in Spanish? Were they English-language learners? What language in particular did they speak? As Otto Santa Ana also stressed in his discussion, the label Latino has political origins that are not grounded in the sociological reality of varied historical origins and cultural and linguistic diversity. More thorough and accurate descriptions of the sample populations could help to interpret the data being collected on samples referred to as Latino and would lead to better understanding of language differences and how to intervene with students.
Bayley agreed with Valdés and colleagues (2009) that longitudinal studies of second-language use by children are needed. Such data might reveal that certain aspects of language may “take care of themselves” with more exposure to language in the school environment, while more explicit and targeted instructional interventions may be needed to see progress in other areas of language. However, a message needs to be sent that language proficiency differs from literacy proficiency. Currently, literacy measures are used for identifying English-language learners for language and learning services, including special education, thereby underestimating the oral-language abilities students possess for learning, further perpetuating achievement gaps.
Bayley also agreed with Valdés and colleagues that defining academic language as distinct from ordinary, everyday language does not seem fully warranted given the linguistic evidence at this point. He likewise agreed that native versus nonnative is not a meaningful way of distinguishing research samples.
Labov’s report linking the absence of certain grammatical forms in African American vernacular to performance on applied mathematics problems suggests the need to create more valid assessments of mathematics and other content areas that are not contaminated with unnecessary linguistic demands. Likewise, assessments are needed to more
accurately distinguish English-language learners and second-dialect speakers from students with learning disabilities, given data reported by Artilles and Ortiz (2002), as well as Rueda and colleagues (Rueda et al., 2002) and others showing that many states use systems that appear to dramatically overrepresent English-language learners in special education programs.
In the future, research is needed to document the language learning trajectories of low-literacy learners as well as English-language learners’ full range of proficiencies, including both first and second language, first and second dialects, and literacy practices used in communities and in school. Some, such as Heath and Kramsch (2004), argue that peer influences are likely to influence how students engage in literacy practices and construe literacy tasks. Thus, understanding the nature of these peer influences has implications for understanding how to support engagement with literacy in the classroom. Bayley concurred with Labov and Hudley (2009) that serious efforts are needed to communicate research-based knowledge about language to the public, educators, and policy makers since little headway seems to have been made despite having accumulated knowledge across decades.
Deeper Understanding of Dialects
Discussant Otto Santa Ana said that teachers would benefit from understanding how home dialects develop; how dialects affect developing other languages, dialects and registers; pedagogical skills for developing literacy from these starting points; and ways that teachers’ linguistic ideologies may be affecting their practices. Santa Ana urged studying verbal expression in institutionalized social practices to inform instruction. That is, although it is important to measure phonology, grammar, and so on, it could be more helpful to address these issues in the context of practices and to encourage teachers’ understanding of how to develop these elements of language in the context of their use. The notion of academic registers could be a meaningful framework for expanding students’ use of language in ways that go beyond concern with teaching and assessing lists of isolated grammatical forms that do not indicate knowledge of actual uses and control of the register in the context of academic learning. Knowing about and using children’s first languages could be helpful to teachers in this regard. From a broader standpoint, Santa Ana argued that attending to the structural aspects of language may be less important than the opportunities children have to learn, which are affected by many factors, including negative perceptions of the learning abilities of children who speak nonstandard English.
Understanding Achievement Gaps
Discussion ensued as to whether achievement gaps are real given that some in the academic community question their existence, saying the notion derives from a deficit perspective. Although most participants acknowledged the empirical reality of achievement gaps, there was some disagreement about the various sources of the gaps and their relative effects on achievement. Consistent with Valdés’ comments, several participants suggested the need to frame questions about the role of language in academic achievement and achievement gaps in a larger framework and to examine the interaction of language in relation to other factors that also affect achievement: poverty, ideology, discrimination, assessment problems, and so on. Remedies for achievement gaps may differ depending on how language relates to these other factors.
Labov agreed that from the point of view of psychology, investigating this complexity is an important question, but that his data from a psycholinguistic perspective show that at the level of grammar, certain features that can be difficult to teach do have cognitive consequences. Yet it is difficult to say right now, he acknowledged, how important intervening with these features of grammar would be to reducing achievement gaps in light of other factors in a learner’s environment. Still, he argued, as anyone who has worked with children in inner-city schools knows, 3rd and 4th graders cannot access education because they do not know the alphabet and how to decode and comprehend what they read: thus, one “path to reducing poverty and inequality is blocked by the fact of reading failure.”
Though acknowledging Labov’s findings, several participants pointed to concerns about children being given extraordinary amounts of language and literacy education unaccompanied by the academic content that could be the vehicle for helping children develop grammar and other language skills. Schleppegrell reiterated that a goal is to prepare teachers to enable every child in the classroom, whatever their language resources, to begin to engage with grade-level content using the approaches illustrated in Schleppegrell (2009) and that have been implanted widely in some countries, including Australia, as noted earlier by Walqui. Bayley noted that debates about the precise role of language is irrelevant as long as children are not getting services that they need because proficiency measures do not go beyond Roger Brown’s morphemes in classifying students as proficient. The result of this misguided approach to measuring student proficiency is that many students do not receive the appropriate language intervention or academic curricula.
Hoff said that a critical question for her is how to engineer environments that provide young children and students with rich language input. Except for the 5 percent or so of children who have a language-learning
impairment, the evidence is very clear that children learn language from input. Before children learn to read, the relevant input is conversation. After they learn to read, the degree to which they read becomes a very important input variable for learning new and rare vocabulary and grammatical forms. This is true for children learning a first or second language, for children learning two languages simultaneously, and for vernacular: African American children learn the African American dialect or standard English depending on what they’re exposed to. So the question for school achievement, Hoff said, is how to ensure children gain access to continual and rich input.
Walter Wolfram questioned the meaning of rich input and the empirical basis for indicating that children are not getting adequate or rich input to learn language since children around the world learn language from varied language models. Hoff responded that if one defines language acquisition as simply knowing a language, then it is indeed the case that all children get sufficient input to know a language. But, realistically, children vary in the levels at which they know their language so that children enter school with different vocabularies, complexities of grammar, and so on, and if one assumes that these variations matter for academic success, then data are accumulating that point to an operational definition of rich linguistic input. For instance, data show that:
The number of different words that mothers use in talking to their children predicts the children’s vocabulary size in spontaneous speech.
The number of words mothers present in talking with their children predicts children’s vocabulary use in spontaneous speech.
The complexity of maternal utterances predicts children’s vocabulary. This effect cannot be due to genetic influences because teachers’ use of complex sentences in the kindergarten classroom also predicts growth in children’s use of complex sentences and comprehension of complex sentences over the kindergarten year.
The frequency with which mothers expand their children’s utterances and the frequency with which mothers ask their children questions predicts the complexity of children’s noun phrases and the acquisition of auxillary verbs.
These are some of the qualities of language input that predict aspects of language development on which children vary, and this variability has been shown to predict success in school in many areas, such as reading achievement.
Valdés stressed that, for her, the question for learning and achievement is whether it is possible to learn subject matter through flawed lan-
guage? She argued that the answer is yes, according to evidence obtained with international university students learning content in a second language. Yet schools in the United States categorize students using literacy assessments, failing to distinguish between oral-language and literacy skills. Although literacy depends on language, the two are known to be different skills learned under different conditions, with literacy learning being more similar to other forms of academic learning. As long as students with a wide range of oral-language skills are all deemed to be perpetual language learners, “closeted away” with other English-language learners even though conversationally and in many other ways they can use English, an achievement gap will persist caused at least in part by children’s lack of opportunity to learn academic content.
Jeff MacSwan noted that two ideas discussed during the workshop permeate language-minority education and special education. First is the notion that students undergo language subtraction and lose proficiency in the first language as a result of learning a second. The latter is a misconception that emanates from standardized testing, which inappropriately labels children as a-lingual or semi-lingual, in his view. For instance, children in his research whose scores on standardized tests ranged from nonproficient through proficient and who could be labeled as a-lingual or semi-lingual on the basis of their test scores were nonetheless empirically indistinguishable in their actual Spanish narration, as indicated in the morphological and other syntactic characteristics of their narratives.
Second, MacSwan said, is that academic language is often discussed as if it is fundamentally different from language used in other contexts. It would not be productive, he said, for researchers to begin relating features of academic language to cognitive ability as if to imply that certain features of academic language are more syntactically complex. A more fruitful way to proceed would be to conceptualize school language as language used in a particular place for a set of purposes and not as having a higher developmental status. Several participants noted that Schleppegrell (2009) has offered one way to start thinking about how this might be done. The researcher’s task then becomes, MacSwan said, to figure out how to apprentice students into using the language of school while engaging them in rich and appropriately complex academic content.