A Dutch project: Assessment for SLN

The development of tasks for the assessment of Sign Language of the Netherlands (SLN) has been part of a larger ongoing bilingual project for deaf children in the Netherlands (Jansma et al., 1997). The overall goal was to develop assessment tasks for SLN in order to be able to assess the SLN proficiency of deaf children and to develop intervention plans if needed. The project aimed to determine if it is possible to develop instruments that can be used to assess deaf children’s language proficiency in SLN.

The development of the tasks to assess the SLN proficiency in deaf children presents a project in progress. The tasks that were developed or are under development provide a more in-depth investigation than a screening.

The 19 deaf students that were involved in the project were aged 4;07 to 8;02. This represents the age of all children over a two-years period. The children were tested more than once in this three-years period.

The tasks under development for assessing SLN proficiency in deaf children focused both on language comprehension and production.

The tasks focus on two aspects of language acquisition: the acquisition of the lexicon and small areas of the morphosyntax. Regarding the lexicon, three tasks have been developed: (1) Receptive Vocabulary Task, (2) Productive Vocabulary Task, and (3) Sign Meaning Extension Task, and for the area of morphosyntax, (4) the Localization Task and (5) Receptive Verb Agreement Task.

(1) Receptive Vocabulary Task: Children are shown a sign on video and “have to choose one out of four pictures that corresponds in meaning with the particular sign. The items cover nouns, verbs, and adjectives” (Jansma et al., 1997, p. 42).

(2) Productive Vocabulary Task: Productive vocabulary was assessed by using the original version of the ‘TAK Productive Vocabulary Test’ (Verhoeven & Vermeer, 1986). The children were shown pictures, and they had to describe it by just using one sign. Through this task nouns, verbs, and adjectives were assessed.

(3) Sign Meaning Extension Task: This task assesses the “deeper lexical knowledge of sign meaning and relations within the semantic network” (Jansma et al., 1997, p. 42). In a clearly structured questionnaire, the children had to describe the meaning of six signs. The used signs differed in terms of being concrete vs. abstract, everyday use versus academic use, and common sense versus specificity.  

(4) Receptive Localization Task: This task assesses the comprehension of localization, which is suggested as an aspect in signed languages that is prerequisite for the comprehension of verb agreement (Lillo-Martin et al., 1985). The children have to answer questions that ask (1) where a specific referent is localized or (2) which point in space is associated with a particular referent.

(5) Receptive Verb Agreement Task: This tasks focuses on verb agreement. Children were presented sentences in SLN that included verb agreement. The children need to match the of the sentence to one of three presented pictures. Verb agreement in this task covers “locative, locative-directional, and locative-directional-orientational verbs” (Jansma et al., 1997, p. 43) as well agreement with one and two arguments.

The authors encountered two problems when they made the decision to develop an assessment for SLN: (1) a lack of basic research on the grammar and the lexicon of SLN, and (2) a lack of research on the signed language acquisition in deaf children of deaf parents. Additionally, there are certain factors that endanger the validity a of signed language test that focuses on lexical proficiency. “The iconicity of a substantial number of signs is one of these factors, extreme lexical variation (both personal and regional) is another” (Jansma et al., 1997, p. 42).

Within the framework of this project, the focus of the tasks were on lexical knowledge and development of localization and verb agreement. Lexical knowledge is regarded as an important factor in the cognitive-academic language development in children and is related to academic success in school. Regarding the last-mentioned area, localization and verb agreement, a test was developed by Lillo-Martin et al. (1985) to assess the acquisition of these sub domains of American Sign Language in deaf children of deaf parents. Within the framework of the project to develop assessment tasks for SLN, the formats of their Verb Agreement Test and Nominal Establishment Comprehension Test were used as examples with regard to whether similar tasks could be developed for SLN.

While developing the receptive and productive vocabulary tasks of the assessment for SLN (Jansma et al., 1997), an issue concerning translating from spoken Dutch into SLN occurred. This issue will be highlighted in connection with other methodological issues that occurred during the development of the receptive vocabulary task. The other tasks were still under development when this paper was published in 1997.

The Receptive Vocabulary Task

Since no other research on assessment for SLN has been developed, the goal of the project needs to investigate if these tasks can be used to obtain information on the development of SLN. Deaf children on two sites, Rotterdam and Voorburg, were tested on a yearly basis. It was hoped that by comparing the results of each child over time, it could be determined which task items best reflected language development. Certain methodological issues were encountered during the process of developing the Receptive Vocabulary Task. The task is based on a test that has been developed for spoken Dutch (TAK; Verhoeven et al., 1986).

The TAK test consisted of 98 items, starting with easy words and ending with the most difficult words. In this first step, the Dutch words were translated into signs. Items that involved pointing to the body or fingerspelling in SLN were excluded.

Since it was expected that the facilitating effect of iconicity of the signs would influence the answer of the children, a pilot was conducted with four deaf children, aged 6, 7, 9, and 10. The correct answers ranged from 84.5 to 87.6%. This task was also shown to three hearing children, aged 7, 8, and 10 who did not have any prior knowledge of SLN. They were shown the signs on video, and then they had to choose the right picture. They obtained high scores ranging from 60.8 to 72.2%. Asked about their way of choosing a picture, the children often named the iconic relationship between the sign and the picture, even if the overall concept was unknown to them.

In order to improve the task, the facilitating effects of iconicity had to be minimized. This was done by including distracters into the list of items. After all original TAK items had been adapted, the task was piloted on five deaf adults who worked in a school for the deaf. At the end, the task consisted of 93 items.

The now adapted version was used to assess the deaf children in Rotterdam. They obtained scores ranging from 47.3 to 70.9% correct answers.

In the next step, the researchers wanted to see if they reduced the facilitating effect of iconicity of this task. Again, the task was administered with four hearing children between 5 and 6 years old. This group of hearing children also did not have prior knowledge with SLN. One of the four children produced statistically significantly more correct answers than the other three children. This raised the question that if this task required knowledge of a signed language (rather than logical reasoning and problem-solving strategies), one would not expect this result. More hearing children will be tested with this task to find out if this child was an exception or if more children use ‘an iconic strategy’ to solve this task.

In order to pay enough attention to the regional variations between Rotterdam and Voorburg, all items were checked with deaf adults before they were tested on the children in Voorburg. The adults should determine if the items of the task are related to regional variations. It turned out that some signs were totally different between Rotterdam and Voorburg. Consequently, the two regional variations of a target were included in the task.

The adapted version was used to assess the deaf children in Rotterdam and in Voorburg. The analysis of the data has not been reported in published literature. The authors state that the entire idea of “assessing, adapting and reassessing is to determine how the items have to be composed and which items really distinguish between more and less proficient signer” (Jansma et al., 1997, p. 46).

It is not possible to determine if this is a norm-referenced or criterion-referenced test, since it reports from work in progress. The same explains the lack of information about the psychometric properties of this tasks. No more up-to-date information as of yet was/is available.

The presented information reports from a project in progress. Therefore, no clear information about the administration of the test and its time spent can be provided. The provided information from the tasks give at least enough information to conclude that the test itself is not too long, and that the analysis is based on a pass/fail response which minimizes time-consuming scoring procedures. The authors state that they decided to use a structured format of a language test in order to minimize the problem of extensive transcription of the language data. No more recent information is available on the project. This test is not available, since it also only reports from tasks under development.

Among the strengths of the assessment of SLN is that it (1) focuses on both language comprehension and production, (2) involves also younger children (younger than aged 6), (3) aims to be used for deaf students be able in order to assess their SLN proficiency and to develop intervention plans if needed, (4) focuses as well on lexical knowledge (a gap among the tests), and (5) the issue of facilitating effect of iconicity has been raised

Among the weaknesses are that (1) no further going information on the development of the tasks is available, (2) therefore no information on the psychometric properties of the tasks is provided.