The development of reading competency is one of the most significant pedagogical achievements during the first few years of schooling. Although most children learn to read successfully when exposed to reading instruction, up to 18% of children experience significant reading difficulty (Shaywitz, 1998). As a group, young children with speech impairment are at risk of reading impairment, with approximately 50% of these children demonstrating poor acquisition of early reading skills (Nathan, Stackhouse, Goulandris, & Snowling, 2004; Larivee & Catts, 1999). A number of variables contribute to reading outcomes for children with speech impairment including co-occurring language impairment, the nature and severity of their speech impairment as well as social and cultural influences. An area of research that has received increasing attention is understanding how access to the underlying sound structure or phonological representations of spoken words stored in long-term memory account for reading difficulties observed in children (Elbro, 1996; Fowler, 1991). Researchers have hypothesised that children with speech impairment may be at increased risk of reading disability due to deficits at the level of phonological representations (Bird, Bishop, & Freeman, 1995). Phonological representation deficits can manifest in poor performance on tasks that require children to think about the sound structure of words. Knowledge about the phonological components of words is commonly referred to as phonological awareness. Identifying and manipulating phonemes within words are examples of phonological awareness skills. Some children with speech impairment perform poorly on phonological awareness measures compared to children without speech difficulties (Bird et al., 1995; Carroll & Snowling, 2004; Rvachew, Ohberg, Grawburg, & Heyding, 2003). As performance on phonological awareness tasks is a strong predictor of early reading ability (Hogan, Catts, & Little, 2005), there is an important need to determine if children with speech impairment who demonstrate poor phonological awareness, have deficits at the level of phonological representations. This thesis reports a series of studies that investigated the relationship between phonological representations, phonological awareness, and word decoding ability in children with moderate to severe speech impairment. A child with complex communication needs (CCN) who used Augmentative and Alternative Communication (AAC) was also examined to determine how the absence of effective articulation skills influences the development of phonological representations. The study employed a longitudinal design to compare the performance of nine children (aged 3:09-5:03 at initial assessment) with moderate to severe speech impairment and 17 children with typical speech development on novel assessment measures designed to determine characteristics of children's phonological representations. The tasks required children to judge the accuracy of spoken multisyllable words and newly learned nonwords. The relationships between performance on these tasks and measures of speech, phonological awareness and early print decoding were also examined. Four assessment trials were implemented at six-monthly intervals over an 18-month period. The first assessment trial was administered approximately 6 to12 months before children commenced school. The fourth trial was administered after children had completed 6 to 12 months of formal education. The child with CCN completed three assessment trials over a period of 16 months. Data analyses revealed that the children with speech impairment had significantly greater difficulty (p<0.01) judging mispronounced multisyllable words compared to their peers with typical speech development. As a group, children with speech impairment also demonstrated inferior performance on the judgment of mispronounced forms of newly learned nonwords (p<0.05). No group differences were observed on the judgment of correctly pronounced real and nonword stimuli. Significant group differences on speech production and phoneme segmentation tasks were identified at each assessment trial. Moderate to high correlations (i.e., r = 0.40 to 0.70) were also observed between performance on the phonological representation tasks and performance on phonological awareness and speech production measures at each trial across the study. Although no significant group differences were observed on the nonword decoding task, 4 of the 9 children with speech impairment could not decode any letters in nonwords (compared to only 1 child without speech impairment) at the final assessment trial when children were 6-years-old. Two children with speech impairment showed superior nonword decoding ability at trial 3 and 4. The within-group variability observed on the nonword decoding task highlighted the heterogeneity of children with speech impairment. The performances of four children with speech impairment with differing types of speech error patterns were analysed to investigate the role of phonological representations in their speech and phonological awareness development. The child with delayed speech development and excellent phonological awareness at trial 1, demonstrated superior phonological awareness and word decoding skills at age 6 years, although his performance on phonological representation tasks was inconsistent across trials. In contrast, a child with delayed development and poor early phonological awareness demonstrated weak performance on phonological representation, phonological awareness, and decoding at each successive assessment trial. The child with a high percentage of inconsistent speech error patterns generally demonstrated poor performance on phonological representation, phonological awareness and decoding measures at each of the 4 assessment trials. The child with consistent and unusual speech error patterns showed increasingly stronger performance on the phonological representation tasks and average performance on phonological awareness but limited word decoding ability at age 6. The 11-year-old girl with CCN, whose speech attempts were limited and unintelligible, demonstrated below average performance on phonological representation tasks, suggesting that an absence of articulatory feedback may negatively influence the development of well-specified phonological representations. This thesis provides evidence for the use of receptive tasks to identify differences in the phonological representations of children with and without speech impairment. The findings also provide support for the link between the representation of phonological information in long-term memory and children's speech production accuracy, phonological awareness and print decoding ability. The variable performance of some children with speech impairment and the child with cerebral palsy demonstrate the need to consider individual characteristics to develop an understanding of how children store and access speech sound information to assist their acquisition of early reading skills.