An Introduction to Applied Linguistics

particularly in high-stakes contexts. These concerns relate to:
• The validity and reliability of the assessment tools that are used.
• Their administrative feasibility and cost effectiveness.
• Teachers’ preparedness to take on an enhanced assessment role.
Research in both general education and language learning contexts suggests that 
using a small number of context-specific assessment tasks results in low levels 
of generalizability, thus severely limiting the inferences that can be made about 
a learner’s ability (Brindley, 2000a). At the same time, considerable variability 
has been identified in teacher-developed assessment tasks (Wigglesworth, 2000), 
making comparability across tasks difficult and potentially diminishing the value 
of information on student learning outcomes. Other reliability issues which have 
been identified include the difficulties of obtaining acceptable levels of agreement 
between raters on the quality of student writing portfolios (Hamp-Lyons, 1996) 
and inconsistencies in teachers’ observations of student performance (Rea-Dickins 
and Gardner, 2000; Brindley, 2001b). These issues of reliability have served to 
highlight some inherent problems related to teacher-conducted assessment and 
rating, not the least of which is teacher expertise. In this regard, researchers have 
pointed out that teachers cannot necessarily be expected to possess the skills 
needed to develop and administer high quality assessments that can be used in 
high stakes situations. For this reason, it has been argued that a serious investment 
in ongoing professional development on the part of educational authorities is 
necessary if teachers are to become the principal agents of assessment (Brindley, 
2001a; Inbar-Lourie, 2008). In terms of practicality, research studies have 
consistently demonstrated that as well as requiring higher levels of skill on the