Introduction

An understanding of language testing is critical for applied linguists and language teachers. In the early 1980s Alan Davies declared: ‘ … language testing has come of age and is now regarded as providing a methodology that is of value throughout applied linguistics and perhaps in core linguistics too’ (Davies, 1982: 152). A failure to design or select appropriate instruments threatens the validity of the research throughout applied linguistics and second language acquisition (Chapelle, 1994, 1999; Read and Chapelle, 2001). Furthermore, careful construct definition followed by the construction and validation of methodologies to collect data aid innovation in research (e.g. Svalberg, 2009). Language testing is intimately concerned with Validation theory, and so helps researchers to focus on making explicit the claims about the meaning of test scores, and the rationales and evidence that can be brought to bear upon the evaluation of these claims. This further strengthens the core values of language testing researchers and practitioners. It is also crucial for teachers to be able to assess the usefulness of tests that their students may

wish to take. Selecting the most appropriate test may be a motivational factor in learning, and preparing learners for tests has been a widespread practice for as long as there have been tests (Latham, 1877: 5-8). There is a danger in using retired test forms as a means of teaching. If learners are to be well served by tests in the ways described in Part II of this volume, and if classroom and dynamic assessment (see Antón) is to play a transformative role in learning, language testing and assessment must be given a more prominent position in teacher training. As well as its central role in research and learning, the use of language testing in the wider

world has become endemic. On the one hand, language tests may be used to protect the public, as in the case of ensuring pilots and air traffic controllers can communicate effectively in English (see Moder and Halleck); on the other they may be used as part of governmental immigration asylum policy (see Kunnan). It is also the responsibility of test providers and researchers to inform test users of the limitations and dangers of the proposed uses, as discussed by contributors such as Davies and Spolsky. As Davies (1997; this volume) argues, it is the normative role of large-scale language testing

that has led to the requirement for a code of ethics and guidelines for practice that establish a professional morality, or the contract that the language testing profession has with the public.

We feel this code should extend to all forms of language assessment. Although it is a profession, language testing is still a young discipline. With the professionalization of language testing has come a growing interest in its history, and the philosophy and principles that underpin practice (Spolsky, 1995; Fulcher, 2010). Young disciplines have a short collective memory, which means that researchers fail to replicate, and do not build upon what is known. This is often exacerbated by postmodern beliefs in the local, temporal nature of co-constructed knowledge. Postmodernism does not cohabit easily with language testing, which by its very nature makes an assumption that knowledge, skills or abilities are stable and can be ‘measured’ or ‘assessed’. It does this in the full knowledge that there is error and uncertainty, and wishes to make the extent of the error and uncertainty transparent. But that does not detract from the fundamentally empirical approach that lies at the heart of a social science that is based on measurement theory as well as language theory. Finally, we should recall a longstanding argument in favour of normative assessment: to intro-

duce meritocracy to decision making, where otherwise the opportunities and advantages the world has to offer would be distributed according to other criteria such as birth or socioeconomic class, to mention but two. The meritocratic motivation is what Spolsky (1995: 16-24) refers to as the ‘Chinese Principle’ (Miyazaki, 1981), which resonates with reforms in society and education in Britain during the nineteenth century (Roach, 1971), and subsequently exported around the world. Thus, despite its scientific foundations, like all educational assessment, language testing is fundamentally concerned with social and distributive justice (Rawls, 2001). Its methods may involve quantification, but its goals and motivations are thoroughly human.