We generally give a language assessment because we have some decisions that need to be made, such as selecting students for a program, placing them at appropriate levels for instructional purposes, assessing their progress and achievement in the program, or assigning grades. We need to be able to justify the decisions we make on the basis of test scores, so that we can be accountable to the stakeholders—the various individuals who will be affected in one way or another by the assessment and by the way we use it. In order to justify using the results of a language assessment for making decisions, we need to provide a rationale for linking students’ performance on the assessment with the intended use, or decision. This rationale is provided by an assessment use argument (AUA). By demonstrating, through argumentation and the collection of supporting evidence, that our assessment is useful for its intended purpose, we provide the justification we need to be accountable to the individuals who are affected by the assessment and the way it is used.

For many years, we have thought of English as an academic subject as a symbolic system, governed by a fixed body of static rules. As English has become a globalized means of self-expression and social-advancement, however, the previous conceptualization of English no longer suffices.
We are aware of the variability of language use in a way we perhaps were not before. We are aware that native speaker norms may no longer be the standard by which language proficiency should be measured. And, we are aware of the fluidity and mutability of English. It not only has the power to transform; it also is itself transformed by its global community of users.
In order to honor these new awarenesses, we need a different theoretical position on language. In this plenary, I will suggest a view of language as a complex, dynamic system. Such a complexity-theory inspired view is a non-symbolic depiction of language, one that features change and variability. Such a view clearly has implications for language testing and for teaching.

The study of English grammar is beset by two kinds of dogmatism (which have more in common than is widely realized). On one side, we have tradition-bound fundamentalists with a stance reminiscent of biblical literalism: they take the rules to be perfectly clear regulative principles laid down centuries ago by unquestionable authorities. On the other, we have theoretical linguists, who correctly take grammatical principles as a topic for discovery rather than allegiance, but often tend toward an "intelligent design" perspective, insisting that the apparent detail and complexity of grammar must be explainable via some simple and elegant abstract structure such as a logico-linguistic engineer might have designed. Both views are toxic for the proper understanding of grammar. One thing The Cambridge Grammar of the English Language has demonstrated, surely, is that Standard English grammar is (as Oscar Wilde's character Algernon Moncrieff says of truth) rarely pure and never simple.
Discussing both varieties of dogma would go beyond the scope of one lecture, so here I focus mostly on the traditionalists, who rely on a system of analysis that has gone essentially unaltered in two hundred years, and forms the basis for all dictionaries and almost all grammar textbooks today despite having deep and vitiating flaws. The deepest errors stem from a longstanding confusion of category (word class) with function (grammatical or semantic relation). One symptom is the various attempts made at defining category membership in terms of function, yielding the familiar story in which nouns are names of things, verbs are action words, adjectives are qualities of things, adverbs are modifiers of non-nouns, and prepositions relate things to other things. None of this works. In the case of adjectives and adverbs, in particular, new lines of argument reveal that although the categories are clearly distinct, the contrast cannot be reduced to differences in function. New evidence shows that the complementarity view (adjectives modify nouns, adverbs modify non-nouns) is entirely untenable.

The advent of computer technology has made it possible to access a large amount of textual resources on the Internet and there is a growing number of research in the filed of Natural Language Processing (NLP) to investigate various aspects of language on computer. In the same vein, the use of corpora has been increasingly popular among linguists as they shift their attention from intuition-based theory constructions to more empirical, probabilistic views of language.
This new trend has also made a significant impact on the nature of Second Language Acquisition (SLA) theory and the design of lexically-oriented language syllabus construction. In the former, the compilation and analysis of L2 learner corpora and computational psycholinguistic models of language will shed more light on the process of L2 acquisition. It is not just the revival of Error Analysis once popular back in 1960s and 70s, but it reveals more complex patterns of acquisition focusing on the frequencies and distributions of different error patterns as well as underuse/overuse phenomena across proficiency. I will show some of the results of my research using a corpus of more than 10,000 Japanese EFL learners' compositions called the JEFLL Corpus. I will also describe the findings obtained from the analysis of a large spoken learner corpus called the NICT JLE Corpus.
Secondly, I will discuss the implications of corpus-based research for English language teaching. I have developed several different language teaching materials and resources, including the NHK TV English conversation program called "100 Go de Start Eikaiwa (Let's start English with 100 Keywords)!", which is the first corpus-based TV English conversation program ever made. I will show how innovative the program was by describing the analysis of the British National Corpus and the making of the program based on such corpus findings.
Finally, I will argue that corpus linguistics is a methodology, and thus it is important for teachers and researchers to know how to "use the tool". I hope to suggest several ways of bridging the gap between corpus linguists and ordinary English teachers in order to improve English language teaching by making it more "data-oriented".