The Century of Language Change in Columbus, OH (CLCC) Project: 20th Century European American English Results

Vowels: Vowel Variation and Sound Change

There are several distinctive, robust, and significant patterns of vowel variation leading to sound change over time in Columbus that occurred over the course of the 20th Century. These patterns are best seen by looking at the full vowel system vowel plots of various Columbus speakers born during the course of the 20th Century, since these systems act as a sort of "time capsule" for documenting these sound changes and patterns of vowel variation. Given this fact, a number of individual speaker vowel systems have been included on this web site.

These vowel plots can be accessed by clicking on either the "Raw Hz Plots" or "Normalized Plots" tabs, above. The Raw Hz plots show the natural formant frequencies for each speaker, while the normalized plots have been mathematically transformed so that direct speaker to speaker comparisons can be made. (This is something that cannot be done with Raw Hz plots given differences in natural format production which result from physiological differences between men and women).

Although the individual plots are the best place to "see" the patterns of variation and change found among Columbus speakers, the diagnosis of these patterns involves making generalizations across all the individual speaker vowel systems. Therefore, we present a brief textual summary of all of the major patterns of vowel variation and sound change one can see by looking across all of the speaker vowel systems below, and then leave it up to the reader to confirm and further explore these patterns visually by looking at the various vowel plots included in the other subsections of this area of the web site. In addition, readers interested in reading a much more detailed discussion of the patterns summarized below should see the CLCC Presentations page and also the forthcoming dissertation by David Durian entitled A new perspective on vowel variation across the 20th Century in Columbus, OH. A PDF of Durian's dissertation will eventually be posted here when it is complete.

Select Significant Patterns of Vowel Variation and Sound Change in the 20th Century Columbus English Data

Note that a key containing IPA symbols for all of the vowel class key words used in the discussion below is available under the "Vowel Plot Conventions" tab.
The merger of /ɔɹ/ and /oɹ/, so that both are produced as /oɹ/ (displayed as BOARD in the plots). This pattern of merger was mostly complete in the Central Ohio area even by the time the oldest speakers (Generation 1) were born, hence our using BOARD to represent the merged classes in all vowel plots on the site. However, it should be noted that some of our Generation 1 speakers still actually show some residue of the older non-merged contrast in their speech, so that they still produce occasional tokens of words like "horse" or "north" with the /ɔɹ/ vowel rather than the /oɹ/. This means these speaker truly have only a near merger of /ɔɹ/ and /oɹ/ rather than a full merger. However, Generation 2, 3, and 4 speakers pretty much all have a full merger of the sounds.
The realization of a split short-a system among older speakers. Many Generation 1 speakers show the realization of the entire short-a system (represented in our plots as a set of 4 vowel subclasses: BAN, BAT, BAD, and BAG) as a split system akin to that found historically in US East Coast locations such as New York City and Philadelphia. In the split system, /ae/ is realized with a raised nucleus before voiceless fricatives, non-velar nasals, and the voiced stops /b/ and /d/, but then with a non-raised nucleus before other consonants. As time goes on, the system morphs from the split system into a simple nasal system (where only BAN tokens are raised).

However, this change does not happen "over night" and in fact, the transition occurred across several generations in Columbus. The first step in the transition is that the split system first morphs into a more general raising system, where raised /ae/ tokens occur not only before the consonants mentioned earlier, but also some voiced fricatives, as well as /t/. This kind of expanded system is found among many Generation 2 speakers in our data. Following this expansion, the system quickly "evaporates" and "erodes" into a nasal system (where only BAN tokens are raised). A transitional system, where most of the raising is found only with BAN tokens (although there are some "residual" exceptions left over from the older split system style of raising), is found among some Generation 2 speakers and many Generation 3 speakers. Other Generation 3 and most Generation 4 speakers (continue to) show a simple nasal system (only BAN tokens and not other exceptions), which is (today) now the predominate system found among younger Columbus speakers.

The near merger of BOT and BOUGHT. This sound change involves the backwards movement of BOT and the downwards movement of BOUGHT, such that, eventually, the two classes begin to overlap in phonetic space. The degree of overlap varies among speakers, such that some show more or less a full merger of BOT and BOUGHT while others only show partial merger. Merger-like realization appears to be increasing over time--it is found much less often among older speakers than among younger speakers in our data.
Realization of low vowel portion of the chain shift pattern known as the Canadian Shift among younger speakers. The aforementioned backing of BOT, which occurs as a part of the near merger of BOT and BOUGHT, as well as the lowering of /ae/, which results from the transition of the Columbus short a system from split to nasal, also appear to become linked over time in Columbus as a chain shift. As the chain shift solidifies, /ae/ begins to retract, mimicking the backwards movement of BOT. In younger speaker systems, BOT backing and /ae/ backing are quite pervasive.
Parallel backing of the short vowels BET and BIT among some younger speakers. Once /ae/ begins retracting, some younger speakers (Generation 3 and 4) also begin to retract BET and BIT. However, this pattern is not found among all younger speakers, and it appears to be mostly restricted to Generation 3 and 4 white collar males, although some Generation 4 blue collar males have also begun to show this parallel backing tendency. Although BET and BIT retraction are usually discussed in the literature as being linked to /ae/ retraction and BOT retraction, and therefore, an actual part of the chain shift known as the Canadian Shift, our data suggest BET and BIT are not actually a part of that chain shift. This is because not only do not all younger speakers who show BOT and /ae/ retraction show also show BET and BIT retraction, younger white collar women in fact appear to be fronting BIT and BET, even though they are also retracting /ae/ and BOT. Given that younger white collar men are backing BET and BIT while younger white collar women are fronting them, we therefore conclude BET and BIT are actually a parallel shift, and not a part of a chain shift. Thus far, this generalization (that BET and BIT are really a parallel shift that just happens to also co-occur with the low-vowel chain shift in Canadian Shift dialect areas) has only been discussed in Columbus, but we suspect this observation is actually true for all areas showing the Canadian Shift in the US (the US West and US Midland) and Canada.

The fronting of the nuclei of the back diphthongs SHOES, BOOT, and BOAT. The nuclei of SHOES, BOOT, and BOAT are undergoing a fronting tendency among all speakers that is increasing over time. That is, younger speakers show stronger tendencies than older speakers. Given the widespread nature of back diphthong fronting in Columbus, we argue that these changes are related to one another as a parallel shift, and furthermore, that this parallel shift is a paradigmatic case of phonetic analogy (see Schuchardt, 1885 and Vennemann, 1972 for more information on phonetic analogy.)
Fronting, but then later backing, of the nucleus of BOUT. In the vowel systems of Generation 1, 2, and many Generation 3 speakers, the nucleus of BOUT shows a fronting trend. Like the fronting of SHOES, BOOT, and BOAT, this trend increases over time, so that younger speakers show more fronting than older speakers. However, some Generation 3 and many Generation 4 speakers show a reversal of this fronting trend, such that they actually back BOUT rather than continuing to front it like the older speakers. This reversal appears to be linked to the chain shift of the low vowels BOT and /ae/ (discussed above), whereby both BOT and /ae/ retract. As /ae/ in particular begins to back, it eventually begins to "crowd" the phonetic space occupied by BOUT (particularly since BOUT is fronting). Once this occurs, it appears that BOUT then also begins to retract as a response, so as to avoid "colliding" with /ae/ as /ae/ continues to retract.

Raw Hz Vowel Plots

Vowel variation among 20th Century European Americans is best demonstrated via the use of full vowel system speaker vowel plots, as displayed below. The 62 individual speaker vowel plots are divided into the 4 generational speaker cohorts from whom we have data in our study. The plots are then further subdivided into the two social class groups (White Collar or Blue Collar) surveyed in the study, and also by speaker sex. Within each sex group, speakers are ordered according to their birth date.*

Please note that each vowel plot below can be significantly enlarged by clicking on it. When enlarged, each vowel plot opens in a separate browser window, allowing individual speaker plots to be easily compared with one another in more detail. In addition, please note the grey lines in each plot represent each speakers individual grand mean. A key, which defines all of the vowel class key words using IPA symbols, is available under the "Vowel Plot Conventions" tab.

Raw Hz plots, which are unnormalized, are displayed on this page. For normalized results, click on the "Normalized Plots" tab, above. For vowel class key words and other vowel plot conventions, click on the "Vowel Plot Conventions" tab, also above. For recording media, interviewer, and interview date details, click here.

Generation 1 (Born c. 1896-1913)

Blue Collar Females

Blue Collar Males

White Collar Females

White Collar Males

Generation 2 (Born c. 1924-1938)

Blue Collar Females

Blue Collar Males

White Collar Females

White Collar Males

Generation 3 (Born c. 1945-1968)

Blue Collar Females

Blue Collar Males

White Collar Females

White Collar Males

Generation 4 (Born c. 1976-1991)

Blue Collar Females

Blue Collar Males

White Collar Females

White Collar Males

Notes on Recording Sources, Interviewers, and Interview Dates

Unless otherwise noted below, all speakers plotted on this page were interviewed by David Durian either during the pilot phase of this research, in 2004-2005, or during the main phase of the study, in 2008-2009. The 2004-2005 recordings are a mix of cassette and digital audio tape (DAT) recordings (although mostly DAT). The 2008-2009 recordings were all made directly to stereo 44.1 KHz WAV files using a Zoom H2 digital audio recorder.

The exceptional cases not covered by the statements above include all of the speakers discussed below.

"Ann," "Louise," "Arthur," "Frank," "Thea," "William," and "Tobias" were originally interviewed for a series of projects conducted by Ohio Historical Society (OHS) researchers in the 1970s and early 1980s. These recordings are a mix of analog reel-to-reel and cassette recordings, all of which were digitized for use in this project by the OHS in 2009. The recordings of each of these speakers are used by kind permission of the Ohio Historical Society.
"LANCSOH28c" was originally interviewed by Raven McDavid in 1957 for the Linguistic Atlas of the North Central States [LANCS]. The recording used here is a digitized copy of a surviving excerpt from a reel-to-reel recording made at the time of the interview by McDavid. It is used here by kind permission of William Kretzchmar, on behalf of LANCS.
"DAREOH98" was originally interviewed by Dale Huff in 1970 for the Dictionary of American Regional English [DARE]. The recording used here is a digitized copy of the original reel-to-reel recording made at the time of the interview by Huff. It is used here by kind permission of Joan Hall, on behalf of DARE.
"ColumbusMK" was originally interviewed by Erik Thomas in 1984 for a personal research project. The recording used here is a digitized copy of the original cassette recording made by Thomas at the time of interview. It is used here by kind permission of Erik Thomas, on behalf of himself.
"Barbara," "Michelle," "Bill," "Ted," "Terry," "Sheri," "Jason," and "Rachael" were all originally interviewed for the Buckeye Corpus project at OSU in 2000-2001 by various OSU researchers. These recordings were all made to digital audio tape (DAT) in the phonetics laboratory at OSU. The recordings of each of these speakers are used by kind permission of the Buckeye Corpus research team at OSU. For more information on the Buckeye Corpus, see http://www.buckeyecorpus.osu.edu/.

* Special thanks to Josef Fruehwald of the University of Pennsylvania Department of Lingusitics for R scripting assistance which helped allow the vowel plots posted throughout this site to be generated.

Normalized Vowel Plots

Vowel variation among 20th Century European Americans is best demonstrated via the use of full vowel system speaker vowel plots, as displayed below. The 62 individual speaker vowel plots are divided into the 4 generational speaker cohorts from whom we have data in our study. The plots are then further subdivided into the two social class groups (White Collar or Blue Collar) surveyed in the study, and also by speaker sex. Within each sex group, speakers are ordered according to their birth date.*

Please note that each vowel plot below can be significantly enlarged by clicking on it. When enlarged, each vowel plot opens in a separate browser window, allowing individual speaker plots to be easily compared with one another in more detail. In addition, please note the grey lines in each plot represent each speakers individual grand mean. A key, which defines all of the vowel class key words using IPA symbols, is available under the "Vowel Plot Conventions" tab.

Plots normalized using the Lobanov (1971) z-score normalization are displayed on this page. For unnormalized results, click on the "Raw Hz Plots" tab, above. For vowel class key words and other vowel plot conventions, click on the "Vowel Plot Conventions" tab, also above. For recording media, interviewer, and interview date details, click here.