How does the brain learn to read? Investigations combining neuroimaging with artificial orthography learning.

The proposed research will investigate how the brain learns to read. Most words in English have straightforward pronunciations (habitat, grumpy, church). However, some words do not (colonel, chaos, was). There are two main theories for how we can read all these words. One suggests that different brain system represent typical letter-sound rules and exceptional pronunciations. The other argues that all words are read by the same system, and knowledge of word meanings helps us read exception words. I will adjudicate between these ideas by scanning people's brains while they learn to read new words. This will be the first study to look at brain activity during learning. This is vital to understanding how different brain mechanisms help us to learn different types of words. My first study will determine whether different brain regions are active when learning typical versus exception words. In my second study some of the novel words will have a meaning. This will enable me to see whether brain regions that process meaning information are active when learning words with exceptional spellings. This work will help develop models of learning to read. This will improve the quality of future research and contribute to optimising educational practice.

The proposed research will investigate the neural mechanisms that support learning to read aloud. Based on separable predictions made by the dual route (DRC - Coltheart et al., 2001) and triangle (Plaut et al., 1996) models, the research will address three questions: First, does reading aloud depend on separable lexical and sublexical orthography-phonology (O-P) representations, as suggested by the DRC? Second, does the hippocampus support the development of item specific and generalisation behaviour? Third, is knowledge of word meanings (semantics) important for inconsistent word reading, as postulated by the triangle model? These will be addressed by examining changes in neural activity when learning to read an artificial orthography. This novel combination of methodologies is vital if we are to fully understand the brain mechanisms that support reading development.
Study 1: 24 healthy adults will learn to read novel words written in novel characters, varying in the consistency of their spellings. After training we will assess reading of trained items, discrimination of trained from non-trained items, and reading of a further set of generalisation items. Participants will be scanned throughout this process. Addressing Q1, the DRC predicts that lexical representations should be particularly important in learning inconsistent items and in discriminating trained from non-trained items. Generalisation however, should depend on sublexical activations. Regarding Q2, hippocampal activation should support the development of item specific O-P representations. The DRC predicts that this neural system should not also be involved in generalisation, whereas the triangle model predicts it should be.
Study 2: 24 new participants will undertake the same procedure as in Study 1. However, before entering the scanner and learning to read the orthography, a pre-exposure phase will be completed. Participants will hear each item 5 times. For half of the items, meanings will also be learned in the form of novel pictures. The triangle model predicts that pre-exposure to semantics should increase activation in semantic processing regions such as the MTG and AG when reading inconsistent items. This should be mirrored by a decrease in consistency related activity in O-P processing regions.
Neuroimaging studies of reading largely ignore issues of learning. Theories of reading development do not consider the role of the hippocampus. Many existing findings are difficult to interpret due to non-controlled variables such as phonological familiarity. This novel method of examining orthographic learning whilst insider the scanner should thus provide great scientific advances for reading research and practice.