Professor Pierrehumbert paper published in Journal of Phonetics

Professor Pierrehumbert paper published in Journal of Phonetics

Normally, people recognize words faster and more accurately when they have been used just a short time before. This occurs because hearing a word can activate and update the representation of the word in the listener's mind.

A new study by Professor of Language Modelling Janet Pierrehumbert, recently published in the Journal of Phonetics, investigated the way that dialect differences affect the processing of words. It did this by looking at how quickly and accurately people process words repeated in the same dialect versus in a different dialect. The Journal of Phonetics publishes papers of an experimental or theoretical nature that deal with phonetic aspects of language and linguistic communication processes.

The study used speakers and listeners at Ohio State University, where a relatively standard Midwestern (Midland US dialect) is spoken. Some spoke this dialect natively, whereas others had come to Ohio from Great Northern Cities like Chicago and Detroit, where a rather different dialect is spoken. Speech scientists would expect that hearing a word in one's native dialect would provide the most help in recognizing it a short time later, because people have so much experience with their own dialect. They would expect that Midwestern speakers would be helped most by hearing a word in a Midwestern US dialect, and Northern Cities speakers would be helped most by hearing a word in a Northern Cities dialect.

However, the results were a surprise. Although both dialects were perfectly intelligible to both groups, only words spoken in a Midwestern dialect were remembered well enough to help processing later. This was true even for the Northern Cities listeners. In short, the standard and locally relevant dialect was the only one that was reliably remembered. The results support theories in which 1) recognizing a word does not automatically mean storing it in memory and 2) there is a strong role for social and contextual factors in on-line speech processing.

The paper, 'Variation in the strength of lexical encoding across dialects', co-authored by Cynthia Clopper (Ohio State University) and Terrin Tamati (University Medical Center Groningen, The Netherlands) was published in the Journal of Phonetics Volume 58, September 2016.