Topic 6: Building a structural model for scaling language learning

Formal language teaching and learning is one of the most insufficient and unsuccessful endeavors in education. Some students may reach a level where they can actually use the language they studied for communication, but this has been shown to be mainly due to out-of-school encounters with the language (Verspoor, De Bot and Van Rein, 2011). The lack of in school success is a consequence of the absence of a realistic structural model for ordering language learning objectives.

This lecture/workshop firstly reports on a large-scale, longitudinal project to develop a structured approach to teaching English based on selecting and ordering learning objectives according to their immediate and ubiquitous usefulness. The research has inspired the development of the Common European Framework of Reference for Languages (CEFR) and more recently the Global Scale of English (GSE).

Batches of language tasks, of lexical elements and of grammatical structures for English were submitted in a linked design to judgments of usefulness by about 100 experts and random samples of approximately 500 teachers from a pool of 6000 teachers worldwide. After preliminary data cleaning, these judgments were first evaluated for agreement within both groups and secondly across the two groups. Subsequently compiled data were scaled according to the one-parameter Rasch model. The scale was anchored by links to the Common European Framework of Reference for Languages and data from a compilation of several large written and spoken language corpora.
A total of 1800 language use tasks, 22k lemma (representing 37k word meanings and 350 grammar objectives have been scaled. Deletion rate was around 4.5%. DIF analyses revealed teachers of young learners did not agree with teachers of adult learners and likewise teachers of adult learners weren't able to judge young learner objectives.

The project has yielded rich information for course book and test developers enabling them to design psychometrically scaled learning materials. Learning theory models such as developing the 'proximal zone of development' and 'assessment for learning' can thereby be brought into practice.
Secondly, in a workshop setting the audience will be involved in scaling a set of language performance descriptors to experience in real time how agreement on the difficulty of language tasks reveals a common understanding among the participants.


Freitag/ Friday 24.2.2017; 16:30 - 18:00

Raum / Room

GW2 B2.880


John H.A.L. de Jong (Pearson English / Amsterdam VU University)

Dr. Veronica Benigno (Pearson English, UK)

Sprache/ Language

Englisch/ English