Assessment of word juncture behaviours in the semi-structured connected speech of typically developing children: a pilot of a new method

This research outlines the development of a new protocol for labelling speech sound production at word boundaries in connected speech. Our method includes identification of word juncture behaviours that serve to keep words distinct (open junctures), together with behaviours that bind words together (close junctures) in connected speech. Using story re-tell samples from the Variability in Child Speech (VariCS) corpus, we piloted our new protocol with 10 Scottish primary school children with typical speech, aged 6;10 – 7;8 years. For each story re-tell sample (containing, on average, 148.4-word tokens), it took two trained raters, on average, 64.5 minutes to phonetically transcribe and label word juncture behaviours. Intraand inter-rater reliability with labelling was generally high. In this small pilot, the occurrence of open versus close junctures was 26% versus 74% for each story retell (average across samples). We conclude the protocol shows promise as a robust and useful tool for word juncture assessment and larger-scale studies are indicated.