Aspects of semantics and their influence on word production in language impaired and unimpaired individuals

Spoken word production is semantically mediated, but debates remain regarding how the structure and complexity of semantic representations influence the spread of activation at the semantic level and co-activation of other items at the lexical level and how this affects the speed and accuracy of, and brain activity during, word production. This thesis focused on six feature-based semantic variables that capture aspects of semantics (number of semantic features, intercorrelational density, number of near semantic neighbours, semantic similarity, typicality, and distinctiveness) and investigated which of these variables affect picture naming performance. The underlying mechanisms of these variables were explored using a rich methodological approach focusing on different populations (participants with and without aphasia), types of data (behavioural and electrophysiological), and tasks (standard and speeded picture naming). The experimental study reported in Chapter 2 investigated effects of the semantic variables on picture naming in a large group of participants with aphasia. There were effects of number of semantic features, semantic similarity, and typicality on error types, some of which depended on the integrity of the participant's semantic system. A more homogeneous subgroup showed an effect of number of semantic features on naming accuracy. The results were interpreted in the context of current theories of semantics and word production and highlighted that these theories are underspecified regarding the mechanisms by which item-inherent semantic variables might operate. Chapter 3 explored effects of the same semantic variables on picture naming in neurotypical participants. Number of semantic features facilitated performance, while intercorrelational density and distinctiveness had inhibitory effects. These findings were interpreted as being due to spreading activation at the semantic level and competition at the lexical level. In Chapter 4, electrophysiological data collected during overt picture naming was analysed using waveform and microstate analyses. Number of semantic features was significant in the waveform analysis and in the microstate analysis number of semantic features, intercorrelational density, number of near semantic neighbours, and semantic similarity were found to influence activity in the semantic and lexical network involved in word production. This activity is suggested to be either related to the target word itself or distributed across a cohort of co-activated representations. Chapter 5 reports a comparison of effects of semantic variables in speeded deadline and standard picture naming to test whether their effects are systematically stronger in speeded naming. There was a stronger effect of distinctiveness in speeded naming and a stronger effect of number of semantic features in standard naming. These differences could not be explained by greater responsiveness to input in the speeded naming task. Overall, this thesis has resulted in a better understanding of the effects of semantic variables and underlying mechanisms in picture naming. To explain the effects, theories of word production require mechanisms of semantic facilitation and interference, which could be implemented as spreading activation at the semantic level and competition at the lexical level. However, most current theories of word production need further specification to explain these effects.