Replicability of scientific studies grounded on language corpora requires a careful approach of each step from data selection and preprocessing up to significance testing. In this paper, we propose such a replication of a recent study based on a classic conversational corpus (Switchboard). The study (Cohen Priva et al., 2017) focuses on speech rate convergence between speakers in conversation. While the replication confirms the main result of the original study, it also shows interesting variations in the details. Moreover we take this opportunity to test further the study for its robustness with regard to data selection and preprocessing as well as to the underlying model of the variable (speech rate). The analysis also shows that another approach is necessary to consider the complex aspects of the speech rate in conversations. Pushing further a previous analysis is another benefit of replication in general: testing and trengthening the results of other teams and increase validity and visibility of interesting studies and results.
@InProceedings{FUSCONE18.3, author = {Simone Fuscone ,Benoit Favre and Laurent Prévot}, title = {Replicating Speech Rate Convergence Experiments on the Switchboard Corpus}, booktitle = {Proceedings of the Eleventh International Conference on Language Resources and Evaluation (LREC 2018)}, year = {2018}, month = {may}, date = {7-12}, location = {Miyazaki, Japan}, editor = {António Branco and Nicoletta Calzolari and Khalid Choukri}, publisher = {European Language Resources Association (ELRA)}, address = {Paris, France}, isbn = {979-10-95546-21-4}, language = {english} }
Summary of the paper