This report describes two preliminary studies (a pilot study involving 11 children and a main study involving 24 children) that examined letters chosen in the spelling of nonsense words to determine whether children with learning difficulties in spelling exhibit systematic ways of spelling. The studies showed that phoneme-grapheme grammars differ among children with respect to the graphemes used to represent a phoneme--children with specific literacy difficulties are more likely to have graphemes with lower English contingency than children who do not have reading and spelling difficulties. The studies did not provide enough data per child to test theories of individual differences. The next study will aim to collect sufficient data so that half of it can be used to generate a phoneme-grapheme grammar for each particular child and the other half can be used to test that grammar. Three figures are included. Seventeen references, two appendixes describing the nonsense words used in both studies, and a list of 58 reports that may be obtained from the Centre for Information Technology in Education are attached. (KEH)