Solutions for the Assessment of Bilinguals

Virginia C. Mueller Gathercole, Virginia C. Mueller Gathercole

• Publisher: Multilingual Matters
• Year: 2013
• ISBN:
  9781783090167
• Topic: Langues et linguistique
• Subtopic: Linguistique

Solutions for the Assessment of Bilinguals presents innovative solutions for the evaluation of language abilities and proficiency in multilingual speakers – and by extension, the evaluation of their cognitive and academic abilities. This volume brings together researchers working in a variety of bilingual settings to discuss critical matters central to the assessment of bilingual children and adults. The studies include typically developing bilingual children, bilingual children who may be at risk for language impairments, bilingual and multilingual children and adults found in classrooms, and second-language learners in childhood and adulthood. The contributions propose a variety of ways of assessing performance and abilities in the face of the multiple issues that complicate the best interpretation of test performance.

1 Assessment of Bilinguals: Innovative Solutions

Virginia C. Mueller Gathercole

Volume 1 in this series, Issues in the Asessment of Bilinguals, contained a number of studies that argued for changes in the way in which bilingual children and adults are assessed, whether the assessment is of language itself or of cognitive and academic abilities that go beyond language. The authors provided evidence showing clearly that children growing up bilingually are not the same as children growing up monolingually and that even fully fluent bilingual adults perform distinctly from their monolingual peers. The authors in that volume argue that these differences pervade every area of the bilingual’s linguistic knowledge – vocabulary, syntax, etc. – because of the linguistic and sociolinguistic experiences of bilinguals. The upshot of that work is that it is imperative to reformulate the ways in which we assess bilinguals.

The chapters in this volume – Volume 2 – take up this theme and propose some creative and innovative solutions to many of these issues. In some cases, researchers propose novel ways of testing children or adults in their second language (L2) to gain information on their abilities. In other cases, researchers propose ways of developing tools for assessing the first language (L1) of bilinguals and norming these for bilinguals. In other cases, the authors consider whether linguistic features shared by the bilingual’s two languages (in the form of cognates) should be included or excluded from assessments. In still other cases the authors consider modifications to assessment that should take place in contexts in which either adequate information is not available regarding typical development in the L1 or in which changes occur in the patterns of input and, hence, the profiles observed, for bilinguals as they grow older.

We start with chapters that are concerned with the fact that assessment measures do not always exist for both of the languages that a child speaks. One cannot simply take a test developed for one language (e.g. English) and translate it to use for the other language. (This would be akin to testing a piano player with written music scripted for the drums or judging a hockey player’s skating abilities on the basis of scoring criteria established for figure skaters.) These initial chapters examine how one could develop valid tests, either in the L2, or for an L1 for which no tests currently exist. The authors address the fact that the children or adults of concern are bilingual, and they discuss how this fact can affect the structure of the test itself or the participant’s performance in particular aspects of the language.

In the first of these, ‘Identification of Reading Difficulties in Students Schooled in a Second Language’, Fred Genesee, Robert Savage, Caroline Erdos and Corrine Haigh center on the accurate identification of and effective intervention for L2-learning students with reading impairment in school. They specifically address the question of the best approach for assessing students for whom the school language is a second language. These authors provide a thoughtful and thorough review of the factors that might contribute to poor reading in L2 students. They stress that reading abilities may not just have to do with decoding, but with a range of other factors as well, including knowledge of the L2, knowledge of the cultural contexts of texts, and general cognition, including knowledge of the world and inferencing abilities.

They also provide a comprehensive picture of the research to date, suggesting that word-related, language-related and decoding skills relating to the child’s L1 on entering school can be highly predictive of later reading abilities in the L2 and, hence, can provide early information to educators for the identification of which children may be in need of additional support or intervention. They argue, as a result, that one can test L2 children in their L1 (in Kindergarten) to obtain information on their decoding and language abilities (see also Burns, 2013), and this information can be used to predict reading decoding and reading comprehension of the L2 in Grades 1 and 3. In addition, their own data indicate that children may struggle with reading of the L2 for a variety of reasons – some children may have decoding-related difficulties, some may have language-related difficulties and some may have a combination of these. They suggest, therefore, that the best approach calls for assessment of the language abilities as well as the reading abilities of these children, in order to obtain a comprehensive profile of affected students’ strengths and weaknesses. Some students might require intervention that focuses only on reading, while other students may need intervention that includes language as well as reading support.

While a great deal of the research on reading difficulties in children draws on a discrepancy model – wherein specific reading impairments are defined as involving reading difficulties in the absence of other (language, social, cognitive) difficulties – these researchers propose that a good alternative approach for such students is to use a dynamic approach, in which one observes a child’s response to intervention (RtI). By monitoring a child’s improvement over time, RtI can identify reading difficulties, regardless of the source of those difficulties. Such an approach allows for early intervention, in contrast to a ‘wait-and-see’ approach more typically related to a discrepancy model. By carefully monitoring a child’s response to intervention, the practitioner can determine whether the child is progressing as might be expected or whether, instead, he or she needs further intervention. RtI allows early intervention and observation of children whose L1 may not be known by the practitioner. (One additional strength of such an approach is that it precludes having to ‘label’ students who are having difficulty as impai red or learning disabled – arguably one of the factors that contributes to practitioners’ use of the ‘wait-and-see’ approach.)

In the next chapter, ‘What are the Building Blocks for Language Acquisition? Underlying Principles of Assessment for Language Impairment in the Bilingual Context’, Carolyn Letts turns to the issue of developing tests in a young child’s L1 for the purpose of identifying language impairments (LI) in bilingual infants and toddlers. Letts discusses the fact that bilingual children’s development in each of their languages looks different from development in monolinguals, so it is uninstructive, even inappropriate, to assess such children with tests that have been designed for monolinguals in just one of the languages. Children should be assessed in both their languages, but for many languages no tests are available, and we do not know much about normal developmental paths for those languages. Furthermore, we are still far from fully understanding how best to identify language impairments in bilingual children, whatever the language. The best way forward is complicated by a number of facts: (a) bilingual children’s experience with language is very diverse, so it is difficult to draw inferences that apply to all of them; (b) we are lacking norms for bilingual development of language; (c) assessors, whether speech therapists or educational professionals, often lack knowledge of one or other of the child’s languages; and (d) the linguistic performance of bilinguals in one or another of their languages might look quite similar to that of (monolingual) children with language impairment, but their profiles may look that way because of a lack of experience with the language in question, not due to an impairment (see, for example, Paradis, 2010, and commentaries).

Nevertheless, it is imperative to develop, when possible, assessment measures valid for each language in question, and in particular for bilinguals learning those languages. Letts touches on multiple key considerations that a test designer or language assessor needs to pay attention to, or that a speech language therapist (SLT) needs to be aware of, in order to assess bilingual children’s potential language difficulties in any language. First, she emphasizes that, although we now know quite a bit about the likely ‘markers’ for linguistic impairment in English, we still need to discover what the nature of such markers might be for many of the distinct languages in question. What might be an area of difficulty for LI in one language (e.g. tense in English) might not be of relevance for LI in another language. Thus, we need tests designed for the specific languages of interest, based on the phenomena of relevance to the language in question, and preferably normed on bilingual children learning the language.

Second, as a first step towards developing such tests, we can take advantage of some key aspects of language development that appear to be universal – e.g. that one-word utterances evolve before multi-word utterances, that simple sentences emerge before complex sentences and that inferential skills and metalinguistic skills are generally late to emerge. Drawing on such universals, we can begin to search for guides as to how children pass through these stages in the language in question, keeping in mind that bilingual children may typically experience something of a lag in learning particular components of the language.

Third, Letts stresses that, while the possible markers of LI or of specific language impairment (SLI) will differ across languages, some commonalities we might be able to find as factors shared by language-impaired children of all backgrounds include the severity of the delay and difficulty in comprehension. She proposes possible vocabulary measures, elicitation of simple and complex sentences, and the use of spontaneous language samples in any tests developed.

In the following two chapters, work directed precisely at the development of such assessment measures for two particular languages is reported. These have to do with the development of versions of the CDI (Communicative Development Inventories) for two bilingual populations – first, Basque-speaking children, then Irish-speaking children. In both cases, the tests necessarily have to take into account the fact that children are growing up bilingually, since there are no children learning either language who are monolingual; all children learning both languages grow up as bilinguals.

In the case of Basque, María-José Ezeizabarrena, Julia Barnes, Iñaki García, Andoni Barreña and Margareta Almgren, in their chapter entitled ‘Using Parental Report Assessment for Bilingual Preschoolers: The Basque Experience’, discuss their efforts towards the development of a Basque CDI. Since Basque is a morphologically rich language, the CDI had to be adapted and structured to accommodate this richness. These researchers lay out in detail the steps they had to go through to develop a language-relevant version of the CDI for three age ranges, and they report on the general developmental trends for Basque. This includes information on the development of gestures, vocabulary comprehension, vocabulary production, morphology production and morphological complexity between 15 months and 50 months. Recognizing the fact that Basque-speaking children are usually bilingual or multilingual, the authors also discuss the performance of children with varying amounts of exposure to Basque. While differences are not so apparent early on (because of the paucity of data and the low level of performance at initial stages), differences by exposure appear more dramatic with age.

In the subsequent chapter, ‘Using Parent Report to Assess Bilingual Vocabulary Acquisition: A Model from Irish’, Ciara O’Toole proposes a novel way of dealing with the assessment of bilingual children in relation to Irish. While many people have taken a strategy of using two CDIs – one for each language– for children who speak two languages, in the case of Irish, given that most Irish-speaking children are growing up bilingually (in Irish and English), O’Toole has opted to develop a CDI for Irish that compiles information on both of the child’s languages at the same time, in a single instrument. She proposes a bilingual version of the CDI that includes words common to Irish and English (i.e. loan words, cognates) entered as a single item, in contrast to words different in Irish and English, which are marked separately by the parent. She assesses the validity of the measure by comparing the information obtained from parents on this bilingual CDI (collected up to four times across two years per child) with information gleaned from spontaneous speech samples at the same collection points and finds a high correlation in the vocabulary knowledge they attribute to children.

O’Toole argues that the bilingual CDI constructed in this way streamlines the scoring process for parents, and parents do not have to ‘decide’ which language the common word represents. So far, Irish-dominant children have been tested with this measure. With further data, O’Toole plans to develop a version that will be normed according to the home language of the child – only Irish, Irish and English, and only English. This bilingual CDI can provide a good model for the assessment of other bilingual children across the globe.

In the next chapter, ‘Development of Bilingual Semantic Norms: Can Two Be One?’, Elizabeth D. Peña, Lisa M. Bedore and Christine Fiestas similarly focus on the unique challenges faced in relation to the assessment of bilingual children, especially if one wishes to distinguish language-impaired children from normally developing children. These authors focus on how one can develop tests to measure the bilingual child’s proficiency in both languages in tandem. In order to develop parallel tests in the bilinguals’ two languages, they argue, it is essential to follow well-tested psychometric principles. Using such principles, these researchers take us through the steps one can follow to develop such tests – in this case for Spanish–English bilinguals in Texas. They carefully discuss each step, from the choice of items for inclusion in the test to the development of norms that will serve to discriminate children with language impairment from normally developing bilinguals. In the choice of items, it is important to start from both languages (‘dual focus’) and to choose items that are not culturally skewed in favor of one population or the other. The items must also be sensitive to the traits of interest.

With pilot testing and careful analyses, Peña et al. show how one can design tests that are either linguistically equivalent across the two languages or psychometrically sensitive, depending on the goals for the use of the tests. A psychometrically sensitive test will include items that serve to maximally differentiate language-impaired from non-language impaired children, but at the same time yield minimal discrimination between children with different levels of language proficiency (e.g. Spanish-dominant, English-dominant, balanced) and experience. The principles set out by these authors can serve well as a guide for the development of similar valid measures for other bilingual populations.

Hans Stadthagen-González, Virginia C. Mueller Gathercole, Rocío PérezTattam and Feryal Yavas turn to a different question in ‘Vocabulary Assessment of Bilingual Adults: To Cognate or Not to Cognate’. They ask what elements can affect a bilingual’s performance on assessments, and how it is best to approach those elements. They focus on ‘external’ factors, such as level of input, socio-economic factors and the like, as well as factors ‘internal’ to the languages in question and the relationship between them. They investigate the effects of these on performance in Spanish and English vocabulary tests by bilingual, end-state adults in Miami.

With regard to external factors, these authors found that, although all of the adults tested performed at or above the monolingual norm on the two tests, there was evidence of the influence of external factors on performance, such as which language(s) were spoken in the home when the participants were children, which language(s) they spoke in interaction with older siblings and friends, and the socio-economic level of their birth family.

Beyond these, an important question these authors address has to do with one ‘internal’ factor – lexical convergence between the two languages in cognates (or words that have similar forms and meanings in the two languages). Do bilinguals show better performance on cognates than on noncognates in tests of vocabulary comprehension? Better performance would reveal cross-language facilitation either in acquiring these words or in processing these words in such tests.

The data from these adults indeed reveal superior performance in all bilingual groups on cognates over non-cognates. Further examination reveals that if one separates ‘intuitive’ cognates, or English words that even monolingual speakers of Spanish can ‘guess’ the meaning of, from others, the bilinguals perform even better on these than on non-cognates. These results are discussed in terms of their ramifications for the development of language tests for bilinguals and for our interpretation of such tests.

In Chapter 8, Maria Kambanaros and Kleanthes K. Grohmann face a somewhat different challenge. Their chapter, ‘Profiling (Specific) Language Impairment in Bilingual Children: Preliminary Evidence from Cyprus’, addresses assessment difficulties in relation to a community language for which adequate documentation regarding stages of development is lacking. In this case, the target language is Cypriot Greek, which is related to standard Greek, but whose status is debated – is it a separate dialect or a separate language from standard Greek? Regardless, this language variety varies considerably from standard Greek, so that materials available for standard Greek are inappropriate for use by language professionals and teachers wishing to conduct meaningful assessments of children’s oral performance in Cypriot Greek. These authors discuss the development of materials for testing performance, and they report on two studies investigating knowledge of Cypriot Greek in monolingual and bilingual typically developing children and children with SLI. In one study, they examine children’s access to nouns and verbs; in the other, they look at children’s performance in a story-telling task. In the latter, they pay particular attention to the amount of information provided, the number of subordinate clauses used and the length of the longest utterance.

When the performance of typically developing monolinguals, monolinguals with SLI and bilinguals with SLI is compared, these authors report that bilingual children with SLI perform in all respects like the monolingual children with SLI. They thus argue for a ‘delayed’ rather than ‘deviant’ approach for such children, and they conclude that the fact of having two (or more) languages does not lead to greater impairment in bilinguals than in monolinguals. They conclude with some implications of their work for language policy, language assessment and language treatment for bilingual children with SLI.

In the following chapter, it is not acquisition patterns related to a language for which normal patterns of development are unknown that is the issue, but acquisition patterns related to variations in the amount and types of input a child receives in a bilingual community, particularly in relation to the minority language. In ‘Sociolinguistic Influences on the Linguistic Achievement of Bilinguals: Issues for the Assessment of Minority Language Competence’, Enlli Môn Thomas, Virginia C. Mueller Gathercole and Emma K. Hughes explore the factors that influence the performance of Welsh-speaking children in Welsh. Previous research has shown that for both languages of Welsh–English bilinguals, the initial stages of development are directly linked to the amount of input children receive. Those who hear more English than Welsh perform better in English at early stages than those who h...