Reading and Writing: An Interdisciplinary Journal 16: 123–151, 2003.
© 2003 Kluwer Academic Publishers. Printed in the Netherlands.
Levels of phonological representation in skilled reading and in
learning to read
Université Libre de Bruxelles, Bruxelles, Belgium and Universidade Federal de Santa
Catarina, Florianopolis, Brazil
Abstract. This paper refers to and discusses empirical evidence supporting the general
idea that both skilled reading and learning to read capitalize on underlying phonological
representations. These representations must be speciﬁed in terms of degree of abstractness,
units represented and degree of conscious access to these units. In skilled reading, pre-lexical
representations, at different levels of phonological structure, are unconsciously, mandatorily
and automatically activated, in connection with correspondent orthographic representations.
This process is distinct from the intentional and controlled phonological decoding assumed
by the classic dual-route model. Learning to read may be described as the progression from
conscious phonological decoding, in which phoneme awareness plays a critical role, to the
unconscious mapping of orthographic and phonological segments. Phonological dyslexia is
related to some anomaly in speech perception capacities, making it difﬁcult both to develop
efﬁcient conscious segmentation abilities and grapho-phonological decoding. It is suggested
that this conscious speech segmentation deﬁcit is caused by a lack of appropriate segmentation
at the unconscious, perceptual level.
Key words: Dyslexia, Learning to read, Phoneme awareness, Phonological representations,
Skilled reading, Speech segmentation
The idea that the mechanisms of written word recognition, in both begin-
ning and skilled reading, bear some relationship to the structures of spoken
language, especially with phonology, is now largely accepted. However, a
precise speciﬁcation of the levels of phonological representations that are
involved in reading, or that are required to acquire reading ability, is still
lacking. The aim of the present paper is to raise some questions about this
topic and to provide a short review of relevant data in an attempt to answer
Dimensions of phonological representations
A precise speciﬁcation of phonological representations involves a character-
ization of these representations in at least three dimensions: (1) degree of