James E. Purpura , Teachers College, Columbia University

Assessing Grammar

Despite movements to downplay the role of grammar in second and foreign language classrooms, language teachers have always known that grammar provides a fundamental linguistic resource for communication . They have also known that communicative effectiveness depends, to a large extent, on a learner's grammatical knowledge and how this knowledge can be used to understand or produce a range of literal and pragmatic meanings in language use contexts. As a result, grammar and communicative grammar instruction have never ceased to be the mainstay of most second and foreign language classrooms. The importance of grammar in second and foreign language education has been further highlighted by the considerable body of research in SLA (e.g., Doughty & Williams, 1998; Doughty & Long, 2003; Byrd, 2005), showing that language learning is enhanced when grammar instruction is both form and meaning-based and when the development of a learner's explicit and implicit knowledge of grammar is emphasized.

While approaches to teaching grammar have evolved considerably since the 1960's, approaches to grammar assessment have been varied. In some contexts, grammar assessment has remained firmly rooted in structural linguistics, discrete-point measurement, and dichotomous scoring methods of the 1960s (Lado, 1961; Carroll, 1961). In other contexts, grammatical performance is assessed impressionistically by raters scoring one or more components of language use. Still in other contexts, grammar is assessed in two ways: explicitly through a range of selected-response and limited-productions tasks and implicitly through complex language use tasks.

The purpose of this workshop is to question some the assumptions underlying a range of grammar assessment practices. Participants will be presented with a number of grammar assessment tasks. They will be then asked to examine these tasks from a number of perspectives including what construct is being measured (construct definition; construct representativeness), how it is being measured (task design and authenticity); how it is being scored (scoring), and how the assessment might be designed to represent and further learning.

Recommended Reading for the lecture and workshop:

Purpura, J. (2004). Assessing Grammar. Cambridge : Cambridge University Press Chapters 3 and 4.

Purya Baghai, Islamic Azad University, Mashad Branch

Fundamental Measurement with the Rasch Model

Psychometricians have always been concerned about the inadequacy of conventional methods of testing in educational and psychological measurement. The most essential problem with which they have been struggling for decades is the lack of invariance of measuring devices and the dependence of person measures and item measures on tests and populations respectively. The conventional measurement methods in educational and psychological testing, which are still widely practised, are either to count the number of tasks correctly answered by an individual or to report in the form of a percentile score her standing in the population with whom she has taken the test. The problem here is that the index that we get as the indicator of the ability of the person is test and population dependent; with an easier test the individual becomes more able and with a harder test less able; if the population with whom the individual is compared is smart then she obtains a lower percentile, if the population is poor, a higher percentile. The present paper introduces the Rasch model as a theory which solves measurement problems in the human sciences.

Note: The fee for the pre-conference workshops is 25 USD each.

Click here to register for the workshops >>