Comparing instructional methods for teaching Hindi graphs

Comparing instructional methods for teaching Hindi graphs

First Author: Adeetee Bhide -- National University of Singapore
Additional authors/chairs: 
Perfetti, Charles
Keywords: Akshara, Instruction, Orthographic learning, Writing
Abstract / Summary: 

Developing orthographic knowledge in South Asian languages presents a difficult perceptual learning challenge because they have many visually complex graphs. The present study compares the efficacy of 4 methods for teaching Hindi graphs. In Hindi, consonant clusters are separate graphs formed by fusing individual consonantal graphs. Some consonant clusters look similar to their component parts (transparent), whereas others do not (opaque). We taught participants 80 consonant clusters using 4 different methods: 1) copying a graph while the graph and its components are displayed, 2) writing a graph from memory given its components, 3) choosing the correct graph (from several choices) given its components, and 4) choosing the components (from several choices) given the graph. The comparison between methods 1 and 2 examines testing effects, the comparison between methods 2 and 3 examines motor effects, and the comparison between methods 3 and 4 examines directional effects. We measured learning using 4 tests: 1) writing a graph given its pronunciation, 2) reading a graph, 3) hearing a pronunciation and choosing the correct graph, and 4) identifying orthographically legal graphs. Preliminary results show that participants were more accurate on transparent graphs than on opaque graphs on all tests except for the orthographic legality test. Furthermore, participants were more accurate at matching graphs with their pronunciation and identifying orthographically legal graphs when they learned via copying or writing than via choosing the graph or choosing the components. Participants were also slower at matching graphs with their pronunciations when they learned via choosing complex graphs than via choosing components and this difference was most apparent for opaque graphs. After more data collection, we expect to see greater differences among the learning conditions and more interactions between learning condition and transparency.