An Introduction to Applied Linguistics

Figure 15.5 Dimensions defining the options for construct definition
An example of the former would be ChemSPEAK (Douglas and Selinker, 1993), 
a test that requires examinees to perform tasks that use the lexico-grammatical 
constructions from chemistry in order to test ability to speak about chemistry. 
The test does not, however, ask examinees to simulate the performance of ‘doing’ 
chemistry. An example of a ‘general performance’ test would be the oral proficiency 
interview of the American Council for Teachers of Foreign Languages (ACTFL). It 
requires examinees to engage in conversation about themselves and family that 
might come up in a social situation. In contrast to the specific language content 
of the various modules of the Occupational English Test, this test avoids requiring 
the examinee to talk about field specific topics so that the test score can be used 
to indicate capacity for speaking performance in general.
Construct theory is obviously slippery conceptual business, which needs to 
be anchored in the practices of test design and empirical research. Some steps 
toward understanding construct theory from the perspective of test design have 
appeared recently (Read and Chapelle, 2001), but perhaps the most sweeping 
impact on rethinking construct definition is coming from the use of technology 
for developing test methods. Bachman’s (2000) review of the state of the art of 
language testing at the turn of the century included the following observation 
on the contribution of technology in language testing, ‘... the new task formats 
and modes of presentation that multi-media computer-based test administration 
makes possible ... may require us to redefine the very constructs we believe we are 
assessing’ (Bachman, 2000: 9). Today test developers regularly take into account 
the role of technology in the way they define constructs, and the test methods 
they develop (Chapelle and Douglas, 2006). For example, score interpretation 
for a test of writing which requires learners to compose their response at the 
keyboard needs to include the construct of composing at the keyboard. Similarly, 

---

253
Assessment
core interpretation for a test of listening comprehension that provides video and 
audio input for the examinee needs to take into account the learner’s ability to 
productively use the visual channel in addition to the aural one.
Test Methods
Having defined assessment as ‘the act of collecting information and making 
judgements’, we can define test methods as the systematic procedures set out for 
collecting information and making judgements for a particular assessment event. 
Language testers consider test methods as a set of procedures and describe them 
as sets of characteristics rather than by cover-all terms such as ‘multiple-choice’. 
Multiple-choice refers only to one characteristic of a test – the manner by which 
the examinee responds – but any given testing event is composed of a number 
of other factors which should be expected to affect performance. Ideally, the test 
performance (as outlined in Figure 15.1) would be free from any outside influence. 
However, test methods do affect test performance in various ways.
Douglas (1998) theorizes how test methods affect test performance, suggesting a 
series of processes though which the test-taker perceives cues in the test method, 
interprets them and uses them to set goals for task completion, as illustrated in 
Figure 15.6. Consistent with Bachman (1990) and Bachman and Palmer (1996), 
Douglas (1998) suggests that these strategies are key to the manifestation of