NGT-SRT: De ontwikkeling van een zinsherhalingstest voor Nederlandse Gebarentaal

Abstract

As for most signed languages, there is great variation in sign language fluency amongst users of Sign Language of the Netherlands (NGT). These variations depend for example on the age of acquisition of NGT, the type of education followed by the signer, and whether the signer grew up with hearing or deaf family members. Given the lack of suitable language assessment tools available to tap into this variation in NGT fluency in adults (L1/L2), we developed an efficient (short and simple) NGT assessment tool. After exploring the range of existing assessment tools as described by Enns et al. (2016), we developed a sentence repetition task (SRT) for NGT, the SRT-NGT. In line with the existing SRTs (ASL-SRT , Hauser et al., 2008; BSL-SRT , Cormier et al., 2012; DSGS-SRT , Haug et al. 2015), thirty-nine sentences were created, varying in length and complexity. The different levels of complexity were developed conform a combination of guidelines for the existing SRTs in signed languages, guidelines developed by Mayberry (unpublished) for a related test, and guidelines developed in the European COST project, which focused on SRTs in spoken languages: Action IS0804, described by Marinis and Armon-Lotem (2016). The thirty-nine sentences were equally distributed across the three levels of complexity. The SRT-NGT was piloted with a group of L2 learners of NGT, whom were trained to become a sign language teacher or interpreter. As part of their curriculum, their individual sign language communication skills were monitored and scored following the CEFR guidelines (Boers-Visker et al., 2013). Based on their assessment scores, the students are scaled on a level A1 < A2 < B1 < B2 < C in which A1 is the lowest level (poorly proficient) and C the highest (highly proficient). Their performance on the CEFR-scale was used to answer the first research question: (1) “Does the SRT-NGT show sensitivity to tap into different levels of sign language proficiency in line with CEFR scores?”. Further, my aims were to find out if the three pre-determined levels of complexity in the SRT-NGT are indeed observed in the results as expected and whether performance on the SRT-NGT is affected by working memory capacity. The second research question is: (2) “Does the SRT-NGT differentiate between the pre-determined three levels of complexity?” My third research question sounds: (3) “Is there a relation between the SRT-NGT test scores and working memory?” To answer this question a visual digit span test following Wechsler et al. (1997) was administered. In general, we can conclude that the first pilot results are promising. We found a correlation between the participants’ CEFR level and their scores on the test (r (11) = .824, p < .01). Moreover, the SRT-NGT differentiates between the pre-determined levels of complexity. The overall significance effect is (F (2,10) = 69.641, p < .001). Between level 1 and level 2 the significant difference is (F (1,11) = 5.303, p < .05) and between level 2 and level 3 (F (1,11) = 76,112, p < .001). For our third research question, a correlation is found between the participants’ score on the NGT-SRT and their working memory (r (11) = .626, p < .05).