American Sign Language Proficiency Assessment
The American Sign Language Proficiency Assessment (ASL-PA) has been developed within the framework of a larger research project (Maller et al., 1999).
The purpose of the ASL-PA was to develop an assessment instrument that could determine the level of ASL skills for non-native deaf children, with the goal of monitoring their acquisition process. The ASL-PA ought to assess a broad range of linguistic structures of ASL. The selected items were drawn from ASL acquisition studies. In addition, it was ought to be fairly efficient in administration and analysis, showing strong evidence for the psychometric properties. So far, no adequate ASL assessment tool reporting psychometric properties for assessing the ASL proficiency of deaf children has been developed.
The ASL-PA is a screening rather than an instrument for a linguistic, in-depth investigation, because it assesses a broad range of linguistic structures and is rather short.
The ASL-PA is an instrument that is a global skill level assessment that provides no insight where potential problems in the language acquisition process is. It can be used to observe and subsequently monitor the acquisition process of a deaf child over time, but it does not provide a deeper investigation. The ASL-PA was designed for deaf children aged 6-12.
The ASL-PA assesses expressive language skills only. The authors state that in order “to keep the instrument simple, and oriented towards children’s language use, we chose to focus only on morphosyntactic linguistic structures supported by empirical studies of children’s acquisition and use" (Maller et al., 1999, p. 251).
The linguistic structures of 17 child acquisition studies have been reviewed. Eight morphosyntactic structures of ASL were identified. Their order reflect their path of acquisition. The eight linguistic structures were: (1) one-sign/two-sign utterances (single sign, ten different sign, and two sign production), (2) non-manual markers (yes/no questions, wh-questions, topic, conditional), (3) deictic pointing (real world indexing and abstract indexing), (4) referential shifting (role shift and multiple role position), (5) verbs of motion (simple path movement, central object handshape classifier, and secondary object handshape classifiers), (6) aspects and number (aspect, duality, and distribution), (7) verb agreement (verb agreement inflection reveals two features: real world location marked and abstract location marked), and (8) noun-verb pairs (noun-verb pair production and multiple noun-verb production).
Twenty-three target features were identified across all eight linguistic structures, which are based on ASL acquisition studies.
Content validity was drawn from empirical support. The 23 target features were supported by collecting data from four deaf children of deaf parents (DCDP) in three different discourse settings: (1) interview, (2) peer interaction, and (3) story retelling. They were 6, 8, 10, and 12 years old. It was expected that they produce all of the 23 designated target features. The analysis of targets revealed that at least one of the 23 target features was produced by one of the four children in one of the three discourse settings.
Language Sample Elicitation Procedure: The children were engaged and videotaped in three different types of discourse: (1) Interview: the assessor asked the children questions that are likely to elicit or convey a natural conversation, requiring a detailed response. (2) Peer Interaction: two children are engaged to ask each other questions. That sample can be used for both children. (3) Story retelling: children were shown the cartoon ‘The Tortoise and the Hare’.
Participants: 80 deaf children, aged six to twelve. The subjects were divided into three groups based on their linguistic experience, which was determined by their educational setting: (1) native ASL signers/deaf children of deaf parents (DCDP; n=28), (2) deaf children of hearing parents (DCHP), where ASL was used in the educational setting (n=37), and (3) DCHP, where Manually Coded English (MCE) was used in the educational setting (n=15). In setting (3), no deaf adults worked there, and the participants had little exposure to children who know ASL.
Item Analysis: In order to investigate the item/target feature difficulty, an item analysis was run only with the scores of the 28 DCDP. Items were ordered from least to most difficult. Each target feature was subjected to an item analysis to determine which items were most discriminating. Three items were poorly discriminating.
Standard Setting:Three levels of ASL proficiency were established: Level 3: 16 targets or more produced, Level 2: 11 to 16 targets produced, and Level 1: less than 11 targets produced. The reliability coefficient of internal consistency of the 23 target features was high (.81). Inter-rater reliabilitywas established that a second rater coded about 15% of the language data again. The assessment of the test-retest reliability of the ASL-PA is being planned.
Validity: One method to investigate the construct validity of the ASL-PA is to determine whether particular groups of children actually performed how they were expected to perform. Using the method of contrasting groups, children with different linguistic experience were hypothesized to perform significantly differently on the ASL-PA. The results revealed that the three groups with different ASL experiences scored statistically significant different from each other on the ASL-PA.
Investigating concurrent and predictive validity are currently underway. To investigate concurrent validity, each score of the children of the ASL-PA will be compared with the children’s score of another ASL measure. All children were given two subtest of the Test Battery for ASL Morphology and Syntax (T. Supalla et al. 1995, unpublished). The subtest were (1) Verbs of Motion Production (VMP) and (2) Sign Order Comprehension (SOC). These data have not yet been analyzed. It was impossible at the time to report “predictive criterion validity coefficient (correlation of the ASL-PA with educational, psychological, English language, or other ASL proficiency measures). [The authors] stress the importance of obtaining criterion-related validity evidence and encourage such validity studies, as tests become available” (Maller et al., 1999, p. 263).
The ASL-PA is a criterion-referenced test “that is useful for assessing a child’s individual expressive ASL skills against some predetermined level based on language mastery objectives or the child’s own past performance on this scale” (Maller et al., 1999, p. 264).
The ASL-PA is a test that takes only a half hour to administer and about one to two hours to code and score the data for a single deaf child. This is fairly efficient in terms of conducting the test and scoring the elicited language data. When a single target feature occurred once in the language sample of a child (across all three discourse settings), the coder entered the digit ‘1’ in the assigned box of the scoring protocol. If the target feature was not observed, a ‘0’ was entered. The target feature was only scored once, later occurring target features of the same kind were not scored again. A disadvantage is that an assessor needs to have linguistic knowledge of ASL in order to conduct the test. This test is used within the framework of a research project, but is not available to the public since it is still under development. There are planned efforts to use the ASL-PA as a potential assessment in educational settings as well for research purposes. Gathering more language data will enable normative data to be established. That will ensure the appropriateness of the ASL-PA proficiency Levels 1, 2, and 3 and might consider the different language backgrounds and the age of the children (profiles).
Future plans are that once the psychometric properties are satisfactory and its application in educational settings are set, the function and use will be expanded. The authors state that there is a need “for a more ‘diagnostic’ application of the instrument, as opposed to global skill level assessment” (Maller et al., 1999, p. 265).
The recently founded non-profit organization Gloss Institute works on to make the ASL-PA available to teachers (S. Supalla, personal communucation, May 23, 2012).
The strengths of the ASL-PA are that it is (1) a global assessment; it assess globally the child’s expressive ASL proficiency level, (2) items/target features are based on ASL acquisition studies, (3) language samples are elicited from varied discourse contexts, (4) efficient administration and analysis, (5) investigation of psychometric properties, (6) can be used in educational settings serving deaf students, and (7) for research purpose: to be used to research a specific aspect of linguistic structure, e.g. non-manual markers.
Among the weaknesses of the ASL-PA are(1) limited age range: only for deaf children aged six to twelve, (2) knowledgeable tester required: has only been used by researcher with knowledge about ASL linguistic structure, and (3) no large sample norms are presently available.