JCPSLP Vol 15 No 2 2013

Box 1. A multi-step process for assessing bilingual children’s expressive language skills Component of Activity Application to language sample

Additional considerations

comprehensive

analysis (LSA)

assessment

Identify concerns of family and teachers; gather information not available due to biases in traditional testing Determine child’s dominant language (if there is one)

Determine if language sampling will assist with documenting areas of

Use interpreters and cultural

informants to assist with collecting

concern

data

Interviews with parents, teachers, cultural informants, and other stakeholders

Predict child’s ability to complete language sampling tasks in L1 and L2

Look to the literature for validated

parent questionnaires

Use information to assist with diagnostic Interview decisions

Identify child’s familiarity with language testing procedures

Determine which language sampling contexts are most familiar to the child

Consider developmental/

environmental appropriateness of sampling procedures Use caution in interpreting data if no norms are available in L1 Have team members familiar with the language judge the quality of the sample If child has low scores in L2, compare performance to L1 to ensure deficits aren’t simply due to limited English proficiency

Collect and analyse data in dominant language (L1)

Elicit sample in child’s dominant

language

Collect sample from peer who speaks the same language for comparison

Analysis of descriptive and criterion-referenced language data

Collect and analyse data in nondominant language (L2)

Elicit sample in child’s nondominant

language

Document child’s ability to learn language skills, which is a more accurate reflection of language ability when compared to static

Collect baseline data in English using LSA Identify a language sampling Implement intensive intervention focusing context that is meaningful to both

on relative weaknesses within the

the child and his or her functioning in the environment (e.g., related

language sample

Dynamic assessment

assessment

Determine if child showed significant response to intervention (impairment less likely) or if the child continued to struggle (impairment more likely)

to the curriculum)

that weaker English skills are indicative of a true language impairment. Collecting and analysing non- English samples There are many situations where the SLP may be able to obtain a language sample in a language other than English. The examiner may speak the client’s language or the family may elicit the sample under the SLP’s guidance. A final option is to work with a well-trained interpreter, who may also be able to assist with elicitation and transcription of the sample. Heilmann, Miller, Iglesias, Fabiano-Smith, Nockerts, and Andriacchi (2008) showed that by using standardised transcription and coding procedures, separate transcribers who were fluent in the child’s language could achieve strong inter-rater agreement values across two languages (i.e., English and Spanish). The literature should first be reviewed to identify if there is a precedent for transcription rules for that language and if norms exist (e.g., Ooi & Wong, 2012). If there is no guide for the child’s other language in the literature, the general transcription rules can be followed, such as segmentation of utterances and coding of mazes. When limited norms are available, a detailed interpretation of language performance would be inappropriate. Rather, the SLP can refer to the major language milestones in English and look for any substantial deviations from age-level expectations. For example, the SLP could formulate a general interpretation of the child’s mean length of utterance (MLU), which is a key measure that has been found to provide developmental information across multiple languages, including French (Thordardottir

in the comprehensive assessment and as a baseline for dynamic assessment. After collecting the language sample, the recording of the sample will have to be transcribed and coded with the appropriate conventions. Transcription of language samples has been written about extensively (see Miller, Andriacchi, Nockerts, Westerveld & Gillon, 2012 for a review) and tutorials are publically available (e.g., www. saltsoftware.com). Transcripts are typically coded for the presence of inflectional morphemes, which are sensitive to development in Standard English (Brown, 1973). Clinicians can also document lexical and grammatical errors, which are prevalent in children with weak language skills (Heilmann, Miller, & Nockerts, 2010). In addition to microlinguistic features, SLPs are often interested in discourse-level features, such as reduplications and reformulations (i.e., mazes; Miller, Andriacchi et al., 2012), conversational discourse analysis (Damico, 1985), and narrative organisation skills (Westerveld & Gillon, 2010). See Miller, Andriacchi et al. (2012) for a full summary of language sample measures. When reporting English results for a CALD child, the SLP must use caution and explain the results with significant caveats. If the child’s English skills are reported as being considerably weaker than L1, the SLP should not interpret low English measures as being indicative of a language impairment; the low performance on the English sample could simply be a result of limited English proficiency. In this case, the SLP would want to acquire more data from LI and use the English data as baseline in a dynamic assessment. If, however, the child is judged to have English skills that are comparable to L1, the SLP can have greater confidence

89

JCPSLP Volume 15, Number 2 2013

www.speechpathologyaustralia.org.au