Semantic and pragmatic miscues in non-native spoken extended discourse

When non-native speakers are required to produce lengthy discourse in the form of a short monologue or extended turn, there is a greater burden on them to package the utterances into a ‘tighter’ more coherent discourse for the listener. Under these conditions, miscues in specificity and consistency at the semantic level, and relevance at the pragmatic level often accumulate and interact leading to a discourse which can be perceived by the listener as difficult to follow or lacking in coherence. This study analyzes non-native extended discourse produced under such conditions to investigate the nature of these miscues and how they interact with each other at the discourse level. A manipulation experiment was used to test for the psycholinguistic reality of the miscues; an important procedure in discourse analysis if we are to speak with confidence of the validity of the miscues identified on paper. The experiment used a novel technique which allowed for the modification and insertion of features into the spoken discourse as well as their removal. The manipulated discourse was then rated for improvements in coherence against the original sample by a panel of native listeners. Results suggest that these miscues do indeed significantly impair the coherence of the discourse. The patterns of impairment and interaction are discussed in detail.

► This study investigated coherence in non-native spoken discourse.
► Miscues in semantic consistency and pragmatic relevance were identified.
► A manipulation study was employed to repair the discourse.
► Native speakers rated the manipulated discourse as more coherent.
► The miscues have been shown to have psycholinguistic reality.