The relationship between stored phonological representations and speech output

Low quality, imprecise phonological representations have been hypothesized as an underlying deficit in Specific Language Impairment (SLI). This research compared performance on a silent judgement task and a multisyllabic word naming task using the same 10 words, for 21 children with SLI (mean age 7;6), 21 age-matched (AM) (mean age 7;6) and 21 language-matched (LM) (mean age 5;6) peers. The children with SLI demonstrated significantly poorer performance on the judgement task than either AM or LM peers, while performance on the naming task followed a developmental sequence. There was no correlation between the ability to correctly reject inaccurate productions and the ability to correctly name the items. These results support the suggestion of separate input and output phonological representations and that speech output errors should not necessarily be interpreted as indicative of underlying weakness in phonological representations. The research also highlights the value of individually-designed tasks to measure the input phonological representations for specific words.