Statistical Learning in Word Reading and Spelling across Languages and Writing Systems
Rebecca Treiman, Brett Kessler
Abstract
Learning to read and spell involves learning about the written forms of words and how these are linked to language. Writing systems include formal patterns, which pertain to the appearance of written words, and functional patterns, which pertain to links between units of writing and units of language. We review the evidence that learners of a variety of writing systems extract some of these patterns through statistical learning. Statistical learning is often slow and incomplete, however, and direct instruction improves performance. Many studies have focused on statistical-learning skill as an ability of individual learners. We discuss some problems with this approach, arguing that it is more productive to attend to the statistical patterns that writing systems embody. Studies of these patterns and their learning can help explain similarities and differences in learning to read and spell across writing systems.