Computational Models of American Speech

This text offers a probabilistic and rule-based computational model of transcription data. The authors argue that assumptions about the units of spoken language are critical to a computational model. Their models employ suprasegmental elements such as syllable boundaries and stress.