The study posits that mid-sized groups of phonologically trained non-native speakers of English can collect prosodic data that are equivalent to English native-speakers’ annotations. The hypothesis is supported by the results of a classroom experimentinvolving an experimental group of English-proficient Czech (L1) learners annotating prominence and boundaries in English monological texts before and after additional phonological training aided by Rapid Prosody Transcription (RPT). The annotationresults received before the experimental group had the training demonstrate deficiencies of their prosodic annotation occurring under the probable influence of the learners’ mother tongue (Czech). The analysis of disagreements between the experimental group’s and the control group’s (native speakers) annotations demonstrates that non-native listeners rely on slightly different cues when identifying the prosodic structure of an English utterance. Thus, it is concluded that Czech (L1) speakers of English require mandatory annotation practice focused on the differences between their mother tongue and English to perform annotation tasks successfully. The experimental group’s RPT annotations, conducted after a learning intervention, produced much better results and were recognized as statistically equivalent to native speakers’ RPT annotations. The high alignment of the readings obtained by the experimental and control groups on key prosodic parameters demonstrates that crowdsourcing prosodic information from phonologically trained non-native speakers with the help of the RPT method can be employed as an alternative means of validating intonation research when attracting native speakers to research participation is problematic.
Read full abstract