Skip to main content


IGERT Trainee Joseph Fruehwald Studies Language Change in Philadelphians


Josef Fruehwald’s research focuses on language variation and change. His recent research has addressed the relationship between speakers’ fine-grained knowledge of speech sounds (e.g., how to pronounce the vowel ‘ah’ in particular contexts) and speakers’ more abstract knowledge about the speech categories themselves (e.g., whether “ah” actually is a vowel in your language). He is especially interested in how this relationship can be affected by gradual shifts in pronunciation over generations in a speech community. Methodologically, Fruehwald approaches this research by examining how people actually speak in informal settings. He has assisted in the development of automated analysis systems which extract acoustic data from recorded conversations. These systems have been applied to a large body of conversations with everyday Philadelphians recorded over the past 40 years.

In a co-authored paper with William Labov and Ingrid Rosenfelder, they discovered that a number of vowel shifts have changed in a continuous trend over the 20th century, while others moved in one direction until about the 1950s, and then reversed (Figure 1). In his dissertation, entitled The Phonological Influence on Phonetic Change, Fruehwald has found that contrary to conventional wisdom, large categorical effects related to the abstract phonological representation of speech sounds occur very early in the course of sound changes. The conventional wisdom held that changes in pronunciation slowly accumulate, perhaps through speech production errors and speech perception errors until they reach a critical mass – then and only then do language learners ‘reanalyze’ phonological representations associated with the language. What Fruehwald has found is that sound changes occur both where they are not expected (Figure 2) and don’t occur where they are expected (Figure 3). The best theoretical explanation for these patterns is that phonological reanalysis happens first, and the change in pronunciation follows. This analysis has important implications for theories of language change, language acquisition, and phonetic & phonological grammar.

Address Goals

Discovery: Freuhwald’s work is groundbreaking within the speech and language sciences because it weds the ‘old’ (traditional linguistic methods of recording and transcription) with the ‘new’ (automatic extraction of linguistic data from the recordings themselves). This work paves the way for exciting new avenues of research in which large databases of “how we talk” can be built automatically.

Learning: In a number of outreach activities (detailed elsewhere in this report), Freuhwald and his advisor Labov have been engaging the media in discussions of this work. “How we talk” and “how others talk” is something many Americans are interested in, and is tied up with our own identity as an individual and as a culture. This engagement with the media shows the public that this interesting and important topic can be studied scientifically, helping us better understand who we are, who we were, and who we are becoming.