Bilingual children’s Icelandic skills: Standardized language tests and language samples

Abstract

Previous research on bilingualism in Iceland indicates that the gap in Icelandic skills between bilingual children and their monolingual peers is greater than expected based on comparable research on bilingualism in other language communities. We argue that such results prompt extensive investigation on the language status of bilingual children in Iceland using a wide array of assessment tools. The main goal of this investigation was to assess the Icelandic knowledge of bilingual children with different assessment tools and compare the results to normative data for monolingual children. The study targeted sequential bilingual children in the last year of pre-school (5-6 years old) in Iceland. The inclusion criteria were 1) the child was born in Iceland and had started Icelandic pre-school education no later than at 2 to 3 years of age 2) both parents spoke the same language at home (and not Icelandic) 3) the child had not been previously diagnosed with language or neurological disorder. The participants were 25 bilingual children living in two towns in south-west Iceland, Reykjanesbær and Suðurnesjabær. The children were in Icelandic preschool on average 7-8 hours daily and in Icelandic language environments for, on average, about 43 hours a week. The language tests MELB, a newly developed and standardized language test of Icelandic, and the Icelandic-PPVT-4 were administered. MELB measures both language comprehension and spoken language for the components; semantic, syntax, phonology, and morphology. The test was standardized between 2019 and 2020 with 879 children aged 4 to 6 years. The Icelandic version of the PPVT-4 has not been standardized but normative data are available for children in the target age group (Hrafnhildur Ragnarsdóttir, 2018). Spontaneous language samples were analysed and compared to normative values for monolingual children (Jóhanna T. Einarsdóttir & Álfhildur Þorsteinsdóttir, 2015). To estimate the relationship between the language tests and language samples a correlation analysis was conducted. This was the basis for the second objective of the paper which was to investigate whether language samples would provide complementary information to the results from the standardized tests.

The main results show that bilingual children lag significantly behind their monolingual peers in every metric examined. The mean differences were much greater for the language tests than the language samples. The results from the standardized MELB revealed that bilingual participants, on average, were two to three standard deviations below the mean for monolingual peers. Bilingual children showed strengths in phonological skills, but their vocabulary and morphology skills lagged far behind those of their monolingual peers. Scores from the Icelandic translation of the receptive vocabulary test PPVT-4 showed an even greater disparity, with bilingual children’s average score being more than four standard deviations below the monolingual mean. The MLU (mean length of utterance), and TNW (total number of words) and NDW (total number of different words) in spontaneous speech showed a difference under one standard deviation between the monolingual and the bilingual children. The correlation analyses between the language tests and the language samples showed only significantly correlation for the language analyses metric the proportion of grammatical errors and yes/no-answers. This implies that the other language sample measures, MLU, TNW and NDW do add information to the assessment of bilingual preschoolers’ language development of Icelandic.

Several conclusions can be drawn from the results. First, we confirmed the value of a diverse set of assessment tools when evaluating bilingual children’s language development. The bilingual participants had not been diagnosis with Developmental Language Disorder (DLD) and the results indicated that the use of standardized language scores as diagnosis criteria could result in overdiagnosis. Our results therefore suggest that assessment measures such as language samples could provide useful information in addition to other proven methods, such as testing in the home language. The second main conclusion is that the Icelandic language skills of 5-6 years old pre-school children acquiring the majority language as a second language are surprisingly poor. A difference of 3-4 standard deviations on average for children who have been attending Icelandic pre-school for most of their life is below all evidence-based criteria we have found. Indeed, our participants’ results are closer to what could be expected from bilingual children with DLD if we consider research from other language communities. This is unfortunately in line with previous research conducted on other age groups in Iceland. These serious results call for action and changed attitudes in language stimulation for bilingual children in Iceland. It is important is to give these children the opportunity to learn the Icelandic language, which is used in the compulsory schools and further education.