Each year, students in Pacific nations sit high-stakes national and regional examinations of English, their second or other language. The results often determine their secondary schooling choices and trajectories. This paper argues that repeated uninterrupted enactment of these forms of summative language assessment, and a preoccupation with final scores and ranking in Pacific nations including Tonga, has resulted in an imbalance between the dual ‘educational accountability’ and ‘instructional enhancement’ functions of assessment. This current imbalance obscures the powerful formative potential of these assessments, and masks the wealth of information within the scripts themselves for informing pedagogical practice. Failing to ‘listen to’ what scripts can tell us potentially robs education systems, children, teachers, teacher educators and education policymakers of vital real-time feedback for continual responsive improvement and innovation in teaching, learning, and assessment. To address the imbalance, we argue for the value of a multi-disciplinary approach which mines examination scripts for their insights into instructional and assessment improvement. Two small-scale studies are presented as examples of this. Item analysis and error analysis of student answers in a past Tonga Secondary School Entrance Examination (SEE) Class 6 English examination have revealed multiple insights into the nature of test items and test construction as well as students’ productive language abilities and strategies in English as a second language (ESL). The findings from these studies point clearly to the need for continued capacity building in assessment literacy, and the value of placing a solid understanding of the child’s first language and culture at the heart of effective teaching, learning, and assessment of English as a second language.