Conflicting principles are at work in language. For example, talking faster requires less precise articulation, but being clear requires more precise articulation. How are these conflicts resolved? This project studies two competing hypotheses: absolute priority versus compromise. How do humans make these decisions? Again, this project studies two competing hypotheses: all at once or one at a time. Different fields that study language -- linguistics, computer science, and psychology -- take different views of these hypotheses.
The methods used in this project are drawn from all three of these disciplines. The method of typology comes from linguistics: How do different languages resolve conflicts among the same principles? The method of computer modeling comes from computer science: Develop programs that resolve conflicts or learn how conflicts are resolved in some language. The method of naturalistic data analysis comes from developmental psychology: Use data from young children learning English to test and refine the learning model. Two further features of the project are the creation of publicly available computational tools for doing this work, and the recruitment of future linguists by hiring undergraduate research assistants from the economically and ethnically diverse population of students at a large public university.
Whenever we talk, we unconsciously play host to a complex negotiation between conflicting factors. For example, talking faster requires less precise articulation, but being clear requires more precise articulation. This project compared two competing ideas about how this negotiation works. One idea says that stronger factors have absolute priority over weaker ones, so the weaker factor is relevant only in situations where the stronger factor does not care. The other idea says that stronger and weaker factors compromise; the stronger factors have some priority, but the weaker factors are not irrelevant. The first idea is currently the more prevalent view in the field of linguistics, but the second idea is more typical of how priority relationships work in other types of human behavior. In the course of this project, we discovered that the second idea is an adequate and in some ways superior theory of human language. We also showed that a theory that derives the pronunciation of words in a stepwise fashion is superior to a theory that derives pronuncations all at once. We developed and have shared with researchers and students elsewhere some powerful tools for studying these theories with computers. Tangible products of our research include more than 15 articles published or accepted in refereed journals, an edited volume under contract, two doctoral dissertations, and a great many book chapters, conference papers, posters, and other works. A total of 12 doctoral students, 7 of whom are women, received financial support from this grant. Four have completed their degrees and gone on to faculty or post-doctoral positions. The other 8 are making good progress toward their degrees. The project also provided stipends and research experience under the REU program for 17 undergraduates, of whom 12 are women and 2 are members of minority groups.