Data-Driven Learning: Do Learning Tasks and Proficiency Make a
Difference?
Shiauping Tian
(National Taiwan University of Science and Technology)
This paper examines the effectiveness of Data-Driven Learning on learning tasks with different instructional focuses and on students at different proficiency levels. 98 university students
participated in the study. A control/experimental group pre-test/post-test design was implemented. Statistical analysis of the results indicate that students in the DDL group improved significantly more on comparison of word usage and distinctive features of a text type but not on grammar.
Differences in gain scores by students at different levels did not reach a significant level, although the data favored students with higher proficiency on two of the instructional focuses.
INTRODUCTION
Data-Driven Learning is a term coined by Tim Johns (1991) that refers to the learner sorting through large amounts of authentic language data to arrive at patterns or rules of language use. The DDL approach has been widely acclaimed for its potential in language instruction, for it places the learner in the center of the learning process, encourages hypothesis testing and discovery, fosters learner autonomy, and allows the learner to develop important learning skills. Over the years, researchers and language teachers have proposed and implemented DDL approach in the teaching of lexical collocations, affixes, polysemous words, grammatical patterns, and discourse features (Dyck, 1999; Ball, 1998; Tribble, 1997; Kettemann, 1995). A number of studies have also been conducted to provide empirical evidence to the instructional benefits of DDL in language classrooms (Hadley, 2002; Lee & Liou, 2003; Wang, 2002; Someya, 2000; Ilse, 1991; St. John, 2001).
In addition to the instructional benefits of the DDL approach, attention has also been directed to the factors that influence the effectiveness of such an inductive approach. Researchers have discussed the impact of factors such as learner proficiency, learning style, and difficulty of the language pattern on the effectiveness of DDL. In terms of learner proficiency, studies in the past did not seem to agree on who would benefit more from an inductive teaching approach (Carroll, 1964; Shaffer, 1989; Gross, 1991). In a more recent study on the effectiveness of concordancing on learning of collocations, no significant difference was found between learners at different proficiency levels (Wang, 2002).
Another factor is the features of the language rules being taught; in other words, the
complexity of the rules. Researchers have proposed that an inductive approach was more
1. INTRODUCTION With the exponential growth in the scale of Machine Learning and Data Mining (MLDM) problems and increasing sophistication of MLDM techniques...
AND CURRENT LIMITATIONS OF DATA DRIVEN LEARNING ...difference between the predicted density function ...The simple tasks were designed to make a point ...
Transformation-based error-driven learning and natural language processing A case study in Recently, there has been a rebirth of empiricism in the field of...
Data Quality and Semantic Integrity: Learning and Reasoning about Data ...the mediator must know 1) nominal spread means the difference between yield...
Received in revised form 8 January 2012 Accepted 1 March 2012 Available online 16 June 2012 Keywords: Machine learning Ecological data Regression analysis ...
A Brief Survey of Machine Learning Methods for Classification in Networked Data and an Appl_专业资料。Abstract. This paper surveys work from the field of...
A brief survey of machine learning methods for classi?cation in networked data and an application to suspicion scoring Sofus A. Macskassy Fetch Technologies...
A brief survey of machine learning methods for classi?cation in networked data and an application to suspicion scoring Sofus A. Macskassy Fetch Technologies...
热门文档