Читать книгу The Concise Encyclopedia of Applied Linguistics - Carol A. Chapelle - Страница 141

Reading assessment, within the reading‐as‐language view, focuses on estimating linguistic sophistication, rather than promoting the ability to use language for constructing and generating meanings during reading. Linguistic features, such as syntactic complexity and vocabulary density, are treated as key factors that largely determine comprehension obstacles and are used to manipulate item difficulty to construct tests with the desired level of difficulty. From this perspective, reading assessment relies on two layers of deductions—first, gauging reading ability based on observed behaviors elicited, and then inferring linguistic sophistication from the estimated reading ability. For assessment results to be reasonably accurate as an indicator of linguistic sophistication, manipulated features, such as word frequency and structural complexity, must be unambiguously linked to the item difficulty levels. In practice, however, such calibration cannot be conducted solely on the basis of linguistic manipulations. As described above, a number of other factors affect text comprehension. With little regard for nonlanguage‐specific factors, it would be difficult to promote broad‐based reading ability through assessment.

In language‐focused reading assessment, language selection could also constitute a threat to the validity of the intended inferences. Language selection refers to decisions regarding the use of the student's first language or the target language for non‐passage elements (e.g., test administration, task instructions, and comprehension questions). If, for example, the target language is used for task instructions, there is always a likelihood that some learners pursue the task without understanding what they are expected to do. Although a clear grasp of the task, when presented in the target language, could be taken as a partial indicator of linguistic sophistication, score differences derived from non‐passage elements remain unintended, making their interpretation difficult at best and erroneous at worst. It is problematic to determine whether poor performance is attributable to task misconceptions, underdeveloped linguistic knowledge, or inadequate reading skills. Conflation of this sort, arising from unintended linguistic obstacles, constitutes a unique challenge in second language reading assessment.

Lexical scaffolding, such as access to a glossary and dictionary, is another challenge uniquely associated with second language reading assessment. Because gaps between second language vocabulary size and conceptual knowledge vary differently among learners of different ages, with distinct first language backgrounds, and with diverse educational experiences, lexical scaffolding could differently affect their comprehension performance. It is important that the characteristics of the focal group of test takers and the intended inference be considered when deciding about access to lexical scaffolding. If, for example, the objective of reading assessment is to gauge the ability to learn new concepts through reading, scaffolding should reduce the risk of underestimating the ability of interest. If, on the other hand, the primary goal is to estimate the learner's language proficiency, lexical assistance may hamper accurate inference about linguistic sophistication of the learner.

In brief, the reading‐as‐language view currently prevails in language classrooms. Considering the centrality of linguistic knowledge in reading, such a view may seem sensible. However, reading pedagogy under this view rarely promotes the utilization of the learner's cognitive and conceptual resources, and, as a result, gives nominal attention to the additional, perhaps even more critical, role of knowledge of the target language—that is, granting the learner access to her nonlinguistic resources in the first language. Thus, the narrow focus at its worst disregards the most fundamental operation in second language reading. This could pose serious challenges for score interpretation.