Explaining the gap between reading comprehension and reading test theory

In a recent volume chronicling the historical evolution of assessment constructs and the way these are operationalised through language tests, Weir (2013b) noted how for much of the 20th century the teaching and testing of reading focused on lower-level processing to extract factual meaning at the clause and sentence level, rather than on higher-level processing to combine and integrate textbased and readerbased knowledge sources in order to construct a meaning representation for a text, or across a set of texts. By the 1970s, however, the focus was beginning to shift. Weir highlights an editorial published in issue 15 of Reading Research Quarterly (1980:181–182), under the heading ‘Why comprehension?’, in which the editors noted how the earlier focus in reading was giving way to a new emphasis on comprehension. With greater attention being paid to research into comprehension, i.e. exploration of the cognitive processes involved in meaning construction and the skills and strategies involved, the fi eld of reading began to hold greater interest for language teachers and testers than when the focus had been more narrowly limited to the lower-level processes (Urquhart and Weir 1998). Against this background, the last 40 years have witnessed signifi cant expansion in the volume of empirical research conducted in the fi eld of reading assessment. Many of the question formats commonly used in reading tests have been the subject of intense scrutiny with regard to issues of validity. Multiplechoice and cloze, in particular, were the focus of considerable attention during the 1980s and 1990s, with large numbers of studies devoted to analysing the effi ciency of multiplechoice items or the relative merits of one cloze format over another (Alderson 1980, Bachman 1982). Other research began to explore the role of cultural or background knowledge in a reading test (Clapham 1996), the nature of test taker strategies when assessing reading (Cohen 1984) and the value of readingintowriting tasks within an academic study context (Hill and Parry 1992), and this continued into the 21st century. More recently, greater attention has focused on systematically investigating the cognitive processes utilised by test takers during a reading test, in particular how these can be aff ected by the question formats employed (see Khalifa and Weir 2009 for a full discussion of this with an extensive list of references). Expansion has also taken place more broadly in all areas of both fi rst and 1