How to Test Speaking Skills in Japan

43 Chapter 4: Holistic or analytic? Decide your overall approach

Make peace with subjectivity Underhill (1987) points out that one of the key challenges involved with oral tests is their inherently subjective nature Unlike objective exams that contain only one correct answer and are easy to mark reliably, evaluating speaking requires your subjective judgement As a result, consistent levels of reliability are difficult to maintain, thus making it hard to feel confident your scores are accurate and trust worthy In the wide-open world of language testing, it’s certainly pos sible to work around this subjectivity problem by employing vari ous objective marking strategies For example, you could base your grades solely on the amount of words a student speaks in a given amount of time (Breaux, 2016) or make use of an objectified scoring methodology that assigns full or no credit for demonstrably correct or incorrect answers (Madsen, 1983) However, in this book, I’m going to follow Underhill’s (1987: 89) advice to embrace subjectivi ty in all of its messiness in the name of authentic human interaction: The person-to-person aspect is so important in testing oral profi ciency that it cannot be traded away… Face up to the consequent problems of involving human judgement. People are inconsis tent; they don’t always agree, either with each other or with what they said or thought last week. Having made this decision, the oral test designer now finds herself trying to devise a subjective, human testing system with credible claims to reliability. Remember, reliability is critical because it represents the fair ness and consistency that supports the validity of your oral exams Learning to minimize subjective scoring inconsistency is one of the primary skills of successful oral assessment In the end, however, perfection is not the main goal—positive washback that bolsters speaking ability is As long as you take some basic precautions, you will be able to test your students speaking skills with confidence