This paper presents three experiments exploring the perception and production of accents in song. In a perception experiment, participants listened to passages sung and spoken by native and non-native speakers of English. The participants did better at identifying native speakers when listening to the spoken passages. Accents were also judged as more native-like in song than in speech. In addition, two production experiments compared the acoustic characteristics (pitch, duration, F1 and F2) of sung and spoken vowels, produced by native and non-native speakers of English. Both native and non-native speakers changed the pitch and duration of their vowels when singing; the vowel quality was not consistently shifted. Together, the results indicate that the melody imposed by the song impacts the suprasegmental properties of pronunciation whereas the segmental properties remain largely intact. Based on these results, we conclude that a main reason why accents are more difficult to detect in song than in speech is that the rhythm and melody imposed by the song mask intonational cues to accent.

In Cerebral Palsy (CP) children commonly experience semantic deficits. The cause of semantic deficits, and persistence into adulthood, are unknown. The social network hypothesis states decreased frequency of language use in CP individuals leads to impoverished lexico-semantic representations. The embodied cognition hypothesis emphasizes the role of the motor system in processing action concepts. In this study, participants first completed a sentence-reading task while their event related potentials (ERPs) to semantically incongruent sentences were collected. No evidence was found that CP individuals process domain-general semantic incongruence differently to neurotypicals. Participants then made action decisions on lower-limb (kick), and psychological verbs (believe) in an identity-priming paradigm. The ERP component of interest was the N400. There was no evidence to suggest CP individuals have action verb impairments. If the motor system contributes but is not necessary for action semantics, then participants may present no deficits when compared to neurotypicals as shown here.