Multilingual children are a heterogenous group. Simultaneous bilinguals have been learning multiple languages from birth. Sequential bilinguals begin learning a second language after three years of age. Sequential bilingualism can be additive, whereby the introduction of the second language (L2) results in growth in both languages, or subtractive, whereby L2 introduction results in decreasing proficiency in the first language (L1). Proficiency refers to a child’s relative attainment of a given language; in contrast, ability is the capability for language learning (Jacobson & Walden, 2013). Bilingual children can be dominant in one language or can be considered a “balanced bilingual,” with input and output in each language being relatively equal. Most children in the United States become increasingly English dominant as they begin formal schooling (Jacobson & Walden, 2013). 

A Brief Overview of Bilingual Language Development

A silent period is common when a child is first exposed to a second language. During this time, the child focuses on comprehension of the new language and produces little output. The lack of output can be present in both languages. The silent period often lasts from three to six months. Younger children often exhibit longer silent periods than older children. It is important to note that this is part of typical L2 acquisition, and does not indicate a disorder. I saw an example of a preschooler, Jelena, in her silent period when I was observing my SLP graduate students in a Head Start preschool classroom. Jelena did not speak, but she followed directions, nodded or shook her head in response to yes/no questions, and attended to stories and circle-time activities. I once asked her name, and she wrote it neatly, with better handwriting than mine! Because I was only in the Head Start with my students for one semester, I did not have the opportunity to see Jelena as she emerged from the silent period, but I expect that her language development was on track, based on my informal observations.

As noted above, bilingualism can be subtractive, resulting in language loss in L1. When this happens, a child may appear to have low language ability. In these situations, dynamic assessment and information processing tasks, described below, can be particularly helpful as part of the battery of assessment tools used in the differentiation of language difference or language disorder within difference. Language transfer, or interference occurs when a child uses a language characteristic from one language when speaking the other. As children acquire L2, they may use both languages in a conversational turn. For example, a child at the Migrant Head Start preschool for the children of migrant farm workers once said to me, “Courtney! Mira! Una butterfly!” Her use of mira (look) and una (a) were in Spanish, but butterfly was in English.

When evaluating bilingual children, it is important to distinguish between their proficiency in basic interpersonal communication skills (BICS) and their cognitive academic language proficiency (CALP). BICS are the communication skills needed to meet everyday communicative needs, such as social greetings and informal conversations. BICS typically are mastered in two to three years. In contrast, CALP is needed for success in the decontextualized setting of the classroom. CALP provides language for problem-solving, imagining, and reasoning about topics with which the individual has no personal experience. CALP typically takes five to seven years to master (Cummins, 1981). At times, teachers, parents, and others may not realize that the child’s language ability is typically developing, but that the child has not yet had sufficient exposure to English in order to master CALP. Although treatment from a speech-language pathologist would likely not be recommended, it is important that appropriate supports are given in order to facilitate success in the classroom due to incomplete mastery of CALP. A speech-language pathologist (SLP) could have a role in helping teachers to develop effective language supports for children who are still mastering CALP.

The Converging Evidence Approach to the Assessment of Bilingual Children

Castilla-Earls and colleagues (2020) advocate for a converging evidence approach to the assessment of bilingual children. Figure 1 depicts the four methods of assessment that

comprise the converging evidence model.  This approach dovetails with the functional assessment model (Owens, 2012) described in the assessment chapter of this text. The process of assessing the language of child from a culturally or linguistically diverse (CLD) background does not differ greatly from the process of assessing the language of a child from the mainstream culture, but there are important additional considerations when assessing children from CLD backgrounds. 

Parent and Teacher Ratings

Prior to formal evaluation of a child from a CLD background, the SLP must first obtain an evaluation of the child’s classroom performance from the classroom teacher, case history information from the child’s parents, and the results of language proficiency testing (Castilla-Earls et al., 2020; Pieretti & Roseberry-McKibbin, 2016). The Bilingual English and Spanish Assessment (BESA; Pena, Gutierrez-Clellen, Iglesias, Goldstein, & Bedore, 2018) contains two questionnaires that can be used in gathering information from parents and teachers, the Bilingual Input-Output Survey (BIOS) and the Inventory to Assess Language Knowledge (ITALK). The BIOS provides information about the use of each language at home and at school, and the ITALK provides information about the parents’ and teachers’ perceptions of the child’s strengths and areas of concern. The information about language proficiency and dominance determines the best language(s) in which to assess the child. Children with lower English proficiency and dominance in a primary language other than English should be assessed in the primary language. Children with relatively equivalent proficiency in both languages should be assessed in both, because there may have been language loss in L1 and they may still be learning L2 (Pieretti & Roseberry McKibbin, 2016). 

Standardized Measures

Standardized tests can play a role in the assessment of the language ability of children from CLD backgrounds (Castilla-Earls et al., 2020). However, there are a number of considerations when administering standardized tests to children from CLD backgrounds. First, it is ideal to complete testing in both languages. For English-Spanish bilingual children, assessing in either of the two languages has over-identified children as eligible for services (Castilla-Earls et al., 2020). It is acknowledged that completing standardized testing in both languages is not always possible. If testing in one language does not indicate language disorder, testing in the other language may not be need to be formally tested, since language disorder would be present in both languages. However, if one language suggests a language disorder, testing in the other language would be needed to ensure an accurate diagnosis (Peña et al., 2016). Thus, monolingual English-speaking clinicians could begin by testing in English, and if results indicated typical language ability, finding a way to formally test in other languages, such as using interpreters, might not be necessary, as long as the concerns indicated by the referral sources had been addressed.

Most standardized tests do not have versions in multiple languages. Even if a standardized test is available in the child’s other language, it may not be a valid measure of language ability in that language. Often versions of standardized tests in languages other than English are merely translations of the English version, and the items may have less relevance for some cultural groups. Using a translator to translate the English items into another language has the same problem; even if items are presented in the child’s native language, they may not be culturally relevant. For example, when I worked for Migrant Head Start, I administered standardized developmental screenings to Spanish-speaking children. I administered either the English or Spanish version (using a translator for the Spanish version), based on parental preference. I noticed that even the monolingual Spanish-speaking children answered some of the questions in English, such as the question asking for rote counting. When I asked the psychologist about this, she told me that the rote language activities that American parents conduct with their children are not as common in the culture of these migrant families, and many of the children learned these skills at the Head Start, at which instruction was conducted in English, or by watching television in English. This information suggests that these measures may not be culturally relevant for children from some cultural backgrounds.

The Bilingual English-Spanish Assessment (BESA, Pena et al., 2014) is a standard test designed for English-Spanish bilingual children. The BESA Morphosyntax and Semantics subtests have been shown to reliably discriminate bilingual English-Spanish speakers with DLD from those with typically developing language. Adding mean length of utterance in words (MLUw) from a language sample to the BESA increases the diagnostic accuracy (Lazewnik et al., 2019).

A bilingual approach to assessment

A bilingual approach to assessment can used to improve diagnostic accuracy. One method entails testing the child in both languages and using the score in the best language for determining the presence or absence of a disorder. Alternatively, a clinician can test in both languages and average the two scores (Castilla-Earls et al., 2020). A third bilingual approach to standardized testing is conceptual scoring. Although bilingual children tend to earn lower scores than monolingual children on single-language measures of vocabulary, the use of conceptually scored vocabulary measures reveals that word knowledge is a strength of bilingual children (Mancilla-Martinez, et al., 2020). Conceptual scoring of vocabulary measures allows the child to respond in either language, which allows the testing of both languages simultaneously and assesses the knowledge of concepts, rather than language-specific labels. Mancilla-Martinez and colleagues found that the bilingual Spanish-English kindergarteners and second graders in their longitudinal study did use both languages when assessed using the Receptive One-Word Picture Vocabulary Test–4: Spanish-Bilingual Edition (ROWPVT-4: SBE; Martin, 2013b) and the Expressive One-Word Picture Vocabulary Test–4: Spanish- Bilingual Edition (EOWPVT-4: SBE; Martin, 2013a), highlighting the utility of conceptual scoring to capture the word knowledge of these children. Use of Spanish decreased and use of English increased over time, likely due to the use of English at school. On the ROWPVT-4:SBE, children scored higher than the national norm and evidenced a higher rate of growth than the national norm. Expressively, the children’s scores and growth rates were consistent with national norms. These results indicate that, not surprisingly, bilingual children understand more vocabulary words than they produce. Neither receptive nor expressive vocabulary is an area of deficit, and receptive vocabulary is an area of strength. 

Modifying standardized measures

At times, it may be appropriate to make modifications to how standardized tests are scored, although any modifications must be detailed in the SLP’s report, and it must be noted that these may decrease the validity of standardized scores. For example, when I worked in the Orthodox Jewish Yeshivas, I was asked to assess a first-grader’s language ability. At this Yeshiva, reading was first introduced in Hebrew, then in English. At the point in the year at which I administered the CELF-4, the child had had minimal exposure to reading in English. In contrast to reading in English, Hebrew is read right to left. I noticed that on the Concepts and Following Directions subtest, the child was starting at the right for directions such as “the first….” I scored this subtest two ways. My first score was as though the child answered incorrectly the items for which starting at the right or left would make a difference, and my second gave credit if his response was accurate when starting at the right. The first score fell into the low average range, whereas the second score fell solidly in the average range. I explained to the school personnel and parents that this suggested that the child understood the directions, but did not yet realize that English tests start from the left. I had not observed older children at the Yeshiva starting at the right; since they had more exposure to English reading, they recognized that English tests involve starting from the left.

Dynamic assessment

Dynamic assessment, in which a test-teach-test format is commonly used, is useful for working with children from CLD backgrounds. This method of assessment is fluid and responsive, and provides information about modifiability, rather than merely performance at a particular moment in time. Dynamic assessment is discussed in the assessment chapter of this book. More information on dynamic assessment can be found here: https://www.asha.org/practice/multicultural/issues/. A child who learns quickly once given appropriate exposure likely has a difference, but not a disorder; whereas a child who continues to exhibit difficulty with a concept despite multiple learning opportunities may have a disorder within a difference.

Informal measures

Given that many standardized tests are not appropriate for use with children with language differences, it is critical that SLPs employ other assessment techniques in order to obtain an accurate representation of the language ability of children from CLD backgrounds. Information processing measures, assessment of cross-linguistic cognates, assessment of category-sorting flexibility, and structured assessment of morphology have been shown to be useful in the evaluation of the language ability of bilingual children.

Information processing measures, including sentence repetition and nonword repetition have been suggested as clinical markers of language disorder (Conti-Ramsden, 2003, Conti-Ramsden, Botting, & Faragher, 2001; Dollaghan & Campbell, 1998; Gathercole & Baddeley, 1990), and can be useful in differentiating language differences from language disorders (Armon-Lotem & Meir, 2016; Ebert, Kalanek, Cordero, and Kohnert, 2008; Guiberson & Rodriguez, 2013; Lee-James & Washington, 2018; Ortiz, 2021a; Ortiz, 2021b; Pieretti & Roseberry-McKibbin, 2016). The LITMUS Network provides a set of tools, including nonword repetition and sentence repetition tasks, for the assessment of bilingual populations, including speakers of Catalan, Dutch, French, Hebrew, Polish, Russian, and Turkish.

Ebert and colleagues (2008) provide a set of nonwords developed for use with Spanish-speaking children. This task is helpful in diagnosing Spanish-speaking children with DLD in combination with other measures (Guiberson & Rodriguez, 2013; Ortiz, 2021a,b). Nonword tasks also have been developed in Hebrew (Armon-Lotem & Meier, 2016), Dutch (Boerma et al., 2015), French (de Almeida et al., 2017; dos Santos & Ferre, 2018; Thordardottir & Brandeker, 2013; Tuller et al., 2018), and German (Hamann & Abed Ibrahim, 2017; Tuller et al., 2018). When selecting the nonword task, quasi-universal tasks, which have been designed to consider the phonology of multiple languages, have been shown to have greater usefulness in distinguishing difference from disorder than those which are meant to be language-specific, that is, to focus on the phonology of a particular language (Antonijevic-Elliott, et al., 2020; Ortiz, 2021a). When using language-specific nonword tasks, it can be helpful to administer separate tasks designed to represent  the phonology of each language, as low performance on both would be unlikely for a child with typical language ability. Nonword repetition tasks are administered by asking the child to repeat a series of nonwords, and are scored via either percent items correct (number of nonwords correct divided by the total humber of nonword)s or percent phonemes correct (number of correct phonemes produced divided by the total number of phonemes in the nonwords). Some nonword tasks use scoring that counts all substitutions, omissions, and additions as errors, whereas others do not count as incorrect errors which are consistent with phonological patterns noted in the child’s spontaneous speech or those that would be due to influence from the other language spoken by the child.

Cross-linguistic cognates are words that share semantic meaning and phonological overlap between two languages. Assessment of cognates has utility in the evaluation of vocabulary of bilingual children. Squires et al. (2020) revealed that most multilingual children perform better on cognate vocabulary items than non-cognate vocabulary items. They found that most standardized tests do not effectively measure cognate vocabulary, and suggested that testing past the ceiling on vocabulary measures may benefit bilingual children, as a cognate item may occur past the ceiling. False cognates, which are phonologically similar but have different meanings, did not cause particular difficulty for the children in their study. The authors suggested that working on cognates in intervention may be beneficial to bilingual children.

Category sorting flexibility is another measure of semantic language development that is appropriate for use with bilingual children. A category sorting flexibility task, in which bilingual children are asked in both English and Spanish to sort sets of pictures into two groups, then into two different groups (i.e., by a different characteristic than used for the first sorting), and then explain how the pictures were grouped, has been shown to reflect general language development, rather than specific experience in a given language (Simon-Cereijido et al., 2020).

Levey et al. (2020) recommend a cloze procedure to assess derivational and inflectional morphological ability in children whose exposure to and use of English is 40% or higher. In this procedure, the clinician provides a word, such as teach, then asks the child to complete a sentence, such as Mrs. Smith is a ___. 

Language Sample Analysis

Language sample analysis is considered the “gold standard” for language assessment, and is important in the assessessment of the language ability of bilingual children. Word and morpheme omission errors have been found to predict language disorder in bilingual children, whereas measures of lexical diversity may be related more to proficiency than to ability (Jacobson & Walden, 2013). Tense morphology in the primary language and English can be a diagnostic “red flag,” indicating the presence of language disorder in bilingual children (Pieretti & Roseberry-McKibbin, 2016). Grammatical errors per communication unit reliably differentiates bilingual children with and without language disorders (Kapantzoglou et al., 2017).

Assessment of mazes and the change in maze patterns over time also may help to distinguish difference only from disorder with difference in bilingual children. Mazes, or parts of an utterance that do not contribute to the utterance’s meaning, such as repetitions or filled pauses (e.g., um, uh) may result from linguistic uncertainty, which may be caused by linguistic ability or lack of familiarity with a topic. Mazes also may be the result of increased linguistic productivity. As the complexity of utterances increases, mazes may be used as a place-holder. More grammatically complex languages may show evidence of increased maze use in their speakers (Taliancich-Klinger et al., 2021). Because language samples can be repeated frequently, assessing the use of mazes over time and comparing maze use to linguistic complexity can help to determine how and why a child is using mazes.

Systematic Analysis of Language Transcripts (SALT; Miller & Iglesias, 2019) computer software contains bilingual English-Spanish databases against which children’s language samples can be compared. Rojas and Iglesia (2015) provide guidance on transcribing Spanish language samples into SALT. SALT software can automatically compute a number of indices. Further discussion of SALT and language sample analysis is found in the assessment chapter of this book.

In addition to analyzing oral language samples, clinicians should analyze written language samples. These can be administered by the SLP, or the SLP can collect “work samples” from the child’s classroom teacher. Like oral language samples, written language samples can be transcribed and analyzed using computer software. Monitoring a child’s writing throughout the year can be particularly useful in revealing language disorders within differences. For example, Fumera and Wood (2022) revealed similar error patterns on verb tense and verb omissions between English Learners with and without DLD at the beginning of fifth grade, but at the end of fifth grade the English Learners with DLD produced more verb tense and omission errors and lower overall grammatical accuracy than English Learners with typical language ability.

Observation

Observations of the child in multiple contexts are essential to obtain a complete picture of communication strengths and areas for further development. Seeing differences in how the child communicates in a variety of contexts, with a variety of interlocutors, can provide information about when the child is likely to experience difficulty, and what contextual supports could be beneficial. It is vital to have an understanding of cultural norms when conducting observations of children from diverse backgrounds, as the appropriateness of some behaviors differs among cultures. For example, eye contact is expected in American culture, but in some cultures it is considered rude if children make eye contact with adults. Similarly, in American culture we expect children to initiate communication with adults, but in other cultures children are expected to wait until an adult speaks to them prior to speaking.

Summary of Bilingual Assessment

With any assessment technique or combination of techniques, it is critical to ensure that the items are culturally relevant to the child being evaluated. Reaching out to the family and others in the child’s cultural and linguistic community is necessary in order to facilitate culturally appropriate assessment. Parent input is always important in the assessment of language, but it is especially critical in the assessment of bilingual children, as parents will have a sense of how their child’s communication development compares with that of other children from similar backgrounds.

References