Although our society puts a high value on test outcomes, they are often suspect. Garcia and Figueroa (in press) outlined the tests most commonly given to young children in preschool settings. They examined these tests for predictive ability and validity. Among the eleven most widely used tests of school readiness, most have only adequate validity or worse and all have weaknesses of some sort. In spite of the high status of tests in our society, the perception of tests as objective, scientific, and useful is not consistent with the facts regarding language minority children.

Instead, there is the illusion of objectivity. Because the tests are used to predict a child's success or failure in an instructional sequence or program, predictive criterion-referenced validity is important. Yet the tests do not achieve adequate levels of validity. Because tests are used to make decisions about individual placement, the highest standards of technical excellence are required. Because of their psychometric weakness, there are currently no appropriate tests to assess school readiness for children with limited English proficiency (Salvia and Ysseldyke, 1988).

Guidelines for Assessing Bilingual Children

Given the weakness of current assessment procedures and the multifaceted context of learning for culturally and linguistically diverse children, what principles should guide the design of appropriate assessment instruments? We suggest the following: Assessment must be developmentally and culturally appropriate. In addition to taking into account the social and cognitive aspects of development (Bredekamp, 1987), appropriate assessment for language minority children must take into consideration the unique cultural aspects that affect how children learn and relate to other people (Derman-Sparks, and ABC Task Force, 1989). The adult who probes for elaborate speech may elicit culturally appropriate ways of responding rather than test-appropriate ways of answering. Nonverbal cues may be read incorrectly by the child who comes from another cultural background (Lynch and Hanson, 1992). If, as Mehan, Hertweck, and Meihls (1986) claim, the act of testing is a complex social activity, it is imperative to take care to avoid interpretations and prescriptions that are culturally biased and potentially harmful to the individual being assessed.

The child's bilingual linguistic background must be taken into consideration in any authentic assessment of oral language proficiency. Bilingualism is a complex concept and includes individuals with a broad range of speaking, reading, writing, and comprehending abilities in each language. Furthermore, these abilities are constantly in flux. The conditions of language dominance are quickly altered, especially in children who return to their family's country of origin on a regular basis. Furthermore, some bilingual children also code-switch, as is demanded by the social context.

The goal must be to assess the child's language or languages without standardizing performance, allowing children to demonstrate what they can do in their own unique ways. Assessment must be accompanied by a strong professional development component that focuses on the use of narrative reporting, observations of language development, and sampling the child's language abilities. Teachers and staff need to learn what developmentally appropriate outcomes can be expected based on research in first and second language learning. In particular, they need to know the variety of ways in which children develop a second language.
A fully contextual account of the child's language skills requires the involvement of parents and family members, the students themselves, teachers, and staff in providing a detailed picture of the context of language learning and the resources that are available to the child (Nissani, 1990). What is called for is a description of the child's language environment, of the extent to which significant others-adults or children-provide language assistance by modeling, expanding, restating, repeating, questioning, prompting, negotiating meaning, cueing, pausing, praising, and providing visual and other supports. Assessment of the child needs to take into account the entire context in which the child is learning and developing.

Instructionally Embedded Assessment

Because of the limitations of current assessment procedures, there is a growing consensus that the way to assess bilingual children in early childhood education programs is through a portfolio assessment procedure that is developmentally appropriate, linguistically multifaceted, and contextual (Meisels, 1991; Navarrete, Wilde, Nelson, Martinez, and Hargett, 1990; Valdez Pierce and O'Malley, 1992). Such an approach uses performance samples and observational methods to gain a full picture of the child's language abilities and emergent literacy learning.
In order for such an approach to be developmentally appropriate, it must allow for the fact that bilingual language development can follow a number of different paths. A child entering an early childhood education program can have a strong receptive knowledge of a second language (Type 2), can have learned the second language simultaneously with another language and be fairly balanced in both (Type 1), or can have little knowledge of the second language on entering the program (Types 3 and 4). What is developmentally appropriate for one child is not necessarily appropriate for another.

The procedure must also be culturally appropriate in the sense that there is recognition of the cultural differences between bilingual and mainstream children. Latino or Asian children may have learned different ways of interacting with other children than have monolingual English-speaking children. Children from some cultures learn that it is inappropriate to initiate conversations with adults, to engage with other children competitively, or to look directly at adults. Some children require longer "wait times" before they answer an adult's question. Delay or apparent hesitancy in learning new language skills may actually reflect the difficulties bilingual children have in adapting to new cultural ways of interacting.

The method used to assess the bilingual child's language abilities should be informal, based on performance samples and observations (Navarrete et al., 1990). Young children, and especially children from culturally and linguistically diverse backgrounds, have not been socialized to the activity of test taking. Rigidly standardized procedures seriously underestimate a child's capability. Instead, the teacher can use performance samples and observations to make decisions about individual children that are ecologically consistent with the nature of early childhood learning and instruction. The data must be systematic and developmental, so that the teacher knows what progress the child is making (Herrell, 1990).

In addition, because of the current emphasis on emergent literacy in early childhood (Abramson, Seda, and Johnson, 1990; Teale, 1988), it would be beneficial to examine in the context of the child's overall language development, the acquisition of emergent literacy skills, such as knowledge of the functions of written language, emergent storybook reading abilities, writing strategies, and knowledge of letter-sound correspondences (Teale, 1988). By sampling the child's behavior and through structured observations, teachers can begin to develop a picture of the child's growth in various aspects of language and literacy.
In the current school reform effort, assessment and teaching go hand-in-hand (Herrell, 1990). Assessment should be continuous. When the teacher uses assessments that are an integral part of a classroom activity, it becomes possible to see if the child has learned from the activity. Assessment that informs instruction and follows from it is ecologically valid and pedagogically useful. The model of assessment that we are advocating involves a feedback loop in which assessment is "instructionally embedded." Assessment is intrinsically linked to program goals and affects instructional practice. Such a model is consistent with current thinking about assessment and is appropriate for the needs of language minority children.

Program goals that are based on developmentally and culturally appropriate guidelines influence both instructional practice and ongoing assessment. Assessment and instruction are seen to interact. Rather than sitting children down to take one-shot tests, the teacher is constantly observing what her children can and cannot do at different times and in different contexts and adjusting her instruction accordingly. This is what happens normally in early childhood education programs where developmentally appropriate instruction is occurring.

The teacher's running record of the child's growth in each child's portfolio becomes the basis for conferences with parents in which the teacher can document the child's development. The use of authentic assessments will assist the parents in understanding the child's development and how the curriculum furthers that development. Rather than scores on a test that the parents do not understand, the use of instructionally embedded assessment helps parents see what the goals of the early childhood education program are.

In short, current thinking about assessment practices for language minority children leads to the conclusion that assessment should be instructionally embedded. Especially for children from diverse cultural and language backgrounds, the use of scripted, standardized, norm-referenced measures is inappropriate. Observations and performance sampling at different times and in different contexts allow these children to demonstrate their growth and language competencies.