Vowels and Their Technical Presentation in American Speech

Curated by Thomas Purnell

University of Wisconsin-Madison

A cursory glance through recent articles on vowels in American Speech and a visit to talks at the American Dialect Society’s annual meeting might lead budding dialectologists to conclude that in order to study vowels with flair requires the measurement of the midpoint of thousands of vowels uttered in spontaneous speech by hundreds of thousands of speakers and effortlessly produce R-colored vowel space plots. Such thinking, I argue, is misinformed because it isn’t the technology but the application that is the focus. In spite of vowel plots’ ubiquity in modern dialectological analyses, the appearance of the recognizable square diagrams of purported tongue position based on formant values came relatively late to American Speech (2000, then 2003). Don’t be fooled, though. Because the journal did not publish a vowel plot prior to 2000 does not mean that dialectologists are phonetic Luddites. If we reach back toward the journal’s beginnings in 1925, we notice that dialectologists have been reporting scientific measurements — formants, length, and intensity (though not in that order) — for as long as there have been sufficiently accurate technologies to do so. Watching the manner in which the fields of dialectology and speech technology weave back and forth between technological innovation and linguistic discovery, we observe how the history of investigations into vowels begins and ends with vowel length, an oft forgotten aspect of American English vowels. The more important detail to the discussion of vowels is not the internal value of this technological protocol that crosses formant one (F1) against formant two (F2), but that relational aspects of these diagrams support dialectology in critical ways. This is where we see vowel “duration” becoming relevant to vowel space: a vowel is a whole-mouth articulation with a beginning, an end, and a whole lot of verve moving between the two gestures. While some papers in the journal demonstrate a basic method of application, more often than not, this output of technology facilitates discoveries about variation at the individual and group levels. Looking back over the time lapse between when technology arrived in the journal and now, we reformulate old questions. What will technology help us learn about the individual’s behavior relative to other members of their group? Or, how soon before all the phonetic analysis are automated for us and the humans “merely” interpret vowel plots? Stay tuned as recent articles using Mechanical-Turk collected vowels have arrived.

Notes on the Length of Vowels

R-M. S. Heffner
American Speech (1937) 12 (2): 128–134

Kymograph machines were used for sociophonetic research early on, as evidenced by Harriet Lloyd, who published a two-page “Note on Vowel Length” in 1936 and Norman Eliason’s three-page “Two Notes on Vowel and Consonant Quantity” in 1942. Heffner used a hybrid system to simultaneously record voices on a Sound Scriber and kymograph. This allowed him to replay the recordings from aluminum discs to a kymograph for later analysis (described in 1941). Heffner’s paper is selected because it begins a string of six papers that he and his protégés at the University of Wisconsin published on vowel length, noting that there are important variations in our vowels (e.g., that /ɔ/ is not a short vowel in American English).

Source-Language Transfer and Vowel Accommodation in the Patterning of Cherokee English /ai/ and /oi/

Bridget L. Anderson
American Speech (1999) 74 (4): 339–368

Anderson’s paper represents a move forward in two ways. First, the orientation of her paper characterizes a thrust of the mission of American Speech by considering vowels as markers of identity in a contact setting. Second, she comes very close to producing vowel plots; her figures cast normalized vowel fronting against age for Cherokee and Anglo speakers.

The Role of the Individual and Group in Earlier African American English

Walt Wolfram & Dan Beckett
American Speech (2000) 75 (1): 3–33

This is the article that opened the floodgates of vowel plot hell. Well, not really, because it took two years before Valerie Fridland’s paper “Network Strength and the Realization of the Southern Vowel Shift among African Americans in Memphis, Tennessee” (2003) became only the second article to employ vowel plots, with the added trajectories. (More than likely, the real gateway to the Valhalla of Vowels begins with Erik Thomas’s 2001 PADS volume, An Acoustic Analysis of Vowel Variation in New World English.) Importantly, Wolfram and Beckett use vowel plots to examine the heterogeneity of individuals and their use of vowels as representative of identity within a group, a theme returned to in the Fridland piece.

/æ/-Raising in Wisconsin English

Matt Bauer & Frank Parker
American Speech (2008) 83 (4): 403–431

This article — and the preceding one in the same issue, which will go unnamed — may be underread because the technology seems off-putting. Rather, the technology affords dialectologists an opportunity to “fact check” the articulation of /æ/, arguably the most important vowel in the history of American English dialect variation. (NB, contra reports based on vowel space studies, raised /æ/ is not merged with competitors due given tongue shape and articulatory timing differences.) Although ultrasound technology is less available to dialectologists, Bauer successfully hauled his portable gear around Wisconsin like Hanley did for LANE, albeit without the need of a pickup to do so.

African American English in Urban Seattle: Accommodation and Intraspeaker Variation in the Pacific Northwest

Michael Scanlon & Alicia Beckford Wassink
American Speech (2010) 85 (2): 205–224

Scanlon and Wassink take us deeper into the relationship between vowel head and tail through an analysis of a single individual’s vowels. They demonstrate that the African American woman’s trajectories differ in the ethnicity of the interlocutor. What is important here is that one of the important differences is the trajectory in the pin-pen merger, where the vowel qualities are supposed to be short and identical.

Vowel Dynamics in the Southern Vowel Shift

Charlie Farrington, Tyler Kendall & Valerie Fridland
American Speech (2018) 93 (2): 186–222

There are 81 years between Heffner’s paper and this one — coincidentally, the winner of the American Dialect Society’s inaugural Roger Shuy Award for the best paper in American Speech for 2018. What have we learned about vowels in those 81 years? Heffner wrote multiple papers on a feature of American English that is not contrastive but is important for understanding the identity of African American and Cherokee speakers of American English; Farrington, Kendall, and Fridland — and the papers in between that employ vowel plots — confirm that dialect nuances between vowel beginning and end matter.

Bring on the Crowd! Using Online Audio Crowdsourcing for Large-Scale New England Dialectology and Acoustic Sociophonetics

Chaeyoon Kim, Sravana Redy, James Stanford, Ezra Wyschogrod & Jack Grieve
American Speech (2019) 94 (2): 151–194

While not the first American Speech article to use Mechanical Turked data, this article, which focuses on more than just a few vowel relationships in American English, drives us headlong into the analytical abyss: 626 speakers producing 107,079 vowels, acquired by Mechanical Turk, analyzed by the über-analytical tool DARLA, combined with responses from 535 speaker surveys to produce plotted kriging analyses in R. In the end, figure 30 (yes, 30!) demonstrates the persistence of historical boundaries noticed as far back as LANE. Crowdsourcing in this study breathes the methodological air used to produce the voluminous DARE: combine community surveys with individual recordings to get a larger picture of the dialect landscape.