%0 Journal Article %T L2 pragmatics underrepresentation as a washback mechanism in EFL contexts: Linking test-design features to high-stakes test use %J Teaching English Language %I Teaching English Language and Literature Society of Iran (TELLSI) %Z 2538-5488 %A Mirzaei, Azizullah %A Rahimi Domakani, Masoud %A Seyyed Rezaei, Masoumeh %D 2014 %\ 06/30/2014 %V 8 %N 1 %P 1-29 %! L2 pragmatics underrepresentation as a washback mechanism in EFL contexts: Linking test-design features to high-stakes test use %K Construct validity %K washback %K construct-underrepresentation %K L2 pragmatics %R 10.22132/tel.2014.53916 %X Considering the status of 'washback' in validity theory and research, Messick warns that, to establish test validity, one should not rely merely on washback, with all its complexity, but should instead probe test design properties (i.e., authenticity and directness) and test-use characteristics likely to produce or intensify washback. Authenticity ensures that nothing essential is missing in the assessment of the focal construct, or minimal construct-underrepresentation. Similarly, directness deals with minimal construct-irrelevant variance. This study aimed to explore inadequate representation of L2 pragmatics in EFL curriculum and instruction of Iranian high schools as negative washback of construct-underrepresentation of this important dimension of communicative competence in the high-stakes National University Entrance Test (NUET). The participants were 100 EFL teachers, a sample of 220 pre-NUET students, and 50 post-NUET students in northwest, center, and southwest of Iran. A method-triangulation procedure was employed using questionnaires, observations, and semi-structured interviews. The results showed that the test's construct validity is compromised, meaning that NUET is too narrow and deficient, fails to sufficiently sample pragmatic competence as an important facet of 2discrete-point lexico-grammatical knowledge. Further, the findings demonstrated intense negative washback on L2 teaching-learning processes resulting from using NUET scores for drawing high-stakes action inferences in Iran. The findings suggest that, to facilitate optimal positive washback and adequate sampling of the criterion domain in L2 education, NUET must strive to minimize construct-underrepresentation and construct-irrelevant variance in design. %U http://www.teljournal.org/article_53916_9872b3c75b5a054240f683647389d519.pdf