The Influence of Ethics in Language Assessment

Abstract

To understand the varying influence of ethical considerations on language testing, it helps to note the weight assigned to each of the participants involved, namely the test user (usually also owner), the test writer (usually paid by the owner), and the test taker. Each has different interests: The user wants the most efficient way of assigning an interpretable score to test takers and is not constrained by considerations of fairness; the writer is a professional dependent on the test for employment; and the test taker, while personally concerned with fairness, is seldom powerful enough to expect attention to such claims. Historically, political power determined whose interest was dominant. For example, the imperial Chinese examination ignored the desires of the candidates, being satisfied to identify a tiny elite deserving of appointment to office; a similar approach governed tests for the Indian Civil Service or the Prussian magistracy in the 19th century. The high rewards of these tests established the 19th‐century status of testing and its spread to the English and American educational systems. While some mocked and others regretted the growing power of tests, test makers argued successfully in the early 20th century for their effectiveness, ignoring the evidence of those who had argued for the “inevitable uncertainty” of tests and choosing to develop psychometrics as a method of reducing uncertainty rather then paying attention to the uses and misuses of tests. As language tests became ever more powerful in the later 20th century, a number of professional testers insisted on the development and publication of ethical standards, but absence of enforcement procedures and continuing public belief in the infallibility of tests works against attempts to apply ethics.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 19 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.