The “Canadian Shift” in the US Midland: Purely a Chain Shift?
The Canadian Shift in Columbus, Part 2
David DurianThe Ohio State University
The Century of Language Change in Columbus, OH (CLCC) Project
Background
In previous studies, two similar patterns of vowel shift have been reported in various areas located in the US West and Canada:
California Shift Canadian Shift
Background
Both involve the following components:
Close or merged realization of BOT (“short o”) and BOUGHT (“open o”)
Retraction and lowering of the nucleus of BAT ([ae]) Retraction of BET ([E]) Retraction of BIT ([I])
Background
Recent studies suggest California Shift and Canadian Shift are essentially the same shift, minus the outcome form of the completion of the low back merger (Boberg, 2009; Grama & Kennedy, 2009)
In addition, other recent studies have found that some post-1950 born speakers in US Midland Cities are also engaging in Canadian Shift-like realization behaviors
Durian (2008)--Central Ohio--Midland systems show Canadian Shift Bigham (2008)--Southern Illinois--Midland systems show California Shift
Considered together, the findings of these studies suggest: Canadian Shift= California Shift=Midland Shift tendencies described above
Canadian Shift used as term here, since it was the earliest name given to the Shift in the literature
Questions Remaining Unanswered from the Previous Recent Studies
How does the Canadian Shift “work”? Is it purely a chain shift, or is it something else?
If “something else”, what kind of something else?
The findings of Boberg (2005) in Montreal raise the possibility that Canadian Shift may actually have two components: Parallel Shift + Chain Shift
Which direction are BET and BIT moving in the US Midland? Downward? Forward? Some combination?
Observed Possible Trajectories of BET and BIT Discussed in Previous Studies
BIT PUT
BET BUT
BAT BOT/BOUGHT [Canada] or
BOT/BOUGHT [US W/M]
“Classic chain shift” pattern--Primary movement of BET and BIT is downwards “Classic parallel shift” pattern--Primary movement of BET and BIT is backwards
Diagonal pattern--Primary movement of BET and BIT is first backwards, then downwards
Canadian Shift in Columbus
A post-World War II, middle class phenomenon in Columbus
Vowel systems of speakers born before 1945 show sparse evidence of Canadian Shift
Among speakers born after 1945, Canadian Shift tendencies are most robust among men, and appear to be increasing across generational cohorts
That is, younger male speakers (born post-1975) show stronger Canadian Shift tendencies than older males (born 1945-1970)
Analysis of Speaker Systems Ongoing…
Today’s analysis highlights 12 representative normalized speaker vowel systems of middle class speakers from 4 generational cohorts, born during the years 1895-1990
Today’s analysis based on larger data set of 40 speaker vowel systems
Still ongoing--full analysis will ultimately be based on 76 speaker vowel systems (Durian, forthcoming)
All speakers raised in the greater Columbus Metropolitan Area (Urban Core + Suburbs)
Today’s 40 Speaker Population
Generational Cohort White Collar Speakers Blue Collar Speakers Generation 1* (Born 1896-1913)
2 Men 2 Women 2 Men 2 Women
Generation 2 (Born 1924-1937)
2 Men 2 Women 2 Men 2 Women
Generation 3 (Born 1945-1968)
3 Men 3 Women 3 Men 3 Women
Generation 4 (Born 1976-1990)
3 Men 3 Women 3 Men 3 Women
Totals 20 Speakers 20 Speakers
* Includes tape recorded data graciously provided to us by The Linguistic Atlas of the North Central States [LANCS], The Dictionary of American Regional English [DARE], The Ohio Historical Society [OHS], and Erik R. Thomas (North Carolina State University).
Methods
All tokens extracted from casual interview style speech
Typical interview length 60 minutes
10 tokens of each vowel class extracted from each speaker
All vowels measured from tokens occurring in environments which: Did not occur before a nasal or /r/ or /l/ Did not occur following /l/ or /r/ Tokens included in the BAT class also exclude following velars
Data then normalized using the z-score technique (Lobanov, 1971)
Mean of 10 tokens’ normalized values used as plot points in all plots
Vowel Plots
Vowel classes presented as diphthongs use:
A point extracted at 25% of the vowel’s duration to represent nuclei targets
A point extracted at 75% of the vowel’s duration to represent glide targets
Vowel classes presented as monophthongs use:
A point extracted at 50% of the vowels duration to represent nuclei (equivalent to using measurements taken at steady state)
Normalized values are plotted in z-score units (z-score as in standard deviation, not to be confused with “Z” as in BARK units)
Multiple Speaker Comparison Plots
Older Younger
Males
Females
1 2 3 4
Generation
Nucleus Color
BAT, BAN, and BAG in Columbus
As discussed in previous studies (e.g., Thomas, [1989]/1993, 2001; Hartman-Keiser, et al., 1997; Labov, Ash & Boberg, 2006) /ae/ shows raising pre-nasally (BAN) in Columbus
Although not shown in the plots today, all informants (continue to) show some degree of raising for BAN, regardless of generational group
With the exception of speakers showing continuous /ae/ systems, BAG typically shows similar raising or lowering behavior to BAT (see Labov, Ash & Boberg, 2006 for more details)
As with BAN, BAG is not shown in the following plots
The Development of the Canadian Shift in Columbus During the 20th Century
As in Canada and the US West, Canadian Shift in Columbus appears to be initially triggered by close (partial merger) or merged realization of the BOT and BOUGHT classes
In addition, the reversal of the realization of the nucleus of BAT appears contemporaneously in Columbus vowel systems
The Fall and Rise of BAT in Columbus During the 20th Century
BAT in pre-World War II born Columbus speaker vowel systems was quite often realized either as a falling diphthong [Eae] or, less often, as a diphthong showing mild “Southern Breaking” tendencies [aeEae] (Durian, Forthcoming)
Among post-World War II born speakers, BAT becomes increasingly realized as a rising diphthong [aeE], especially among middle class speakers (Durian, Forthcoming)
merged
close- most frequent outcome in our data set
Generation M F
1
2
3
4
Male
Female
Generation M F
1
2
3
4
Male
Female
Generation 1
Generation 4
Generation M F
1
2
3
4
Male
Female
Generation M F
1
2
3
4
Male
Female
Generation M F
1
2
3
4
Male
Female
Generation M F
1
2
3
4
Male
Female
Generation M F
1
2
3
4
Male
Female
Generation M F
1
2
3
4
Male
Female
Generation M F
1
2
3
4
Male
Female
( )
Generation M F
1
2
3
4
Male
Female
Generation M F
1
2
3
4
Male
Female
Generation 1
Generation 4
Generation M F
1
2
3
4
Male
Female
Generation M F
1
2
3
4
Male
Female
Generation M F
1
2
3
4
Male
Female
Generation M F
1
2
3
4
Male
Female
Generation M F
1
2
3
4
Male
Female
Generation M F
1
2
3
4
Male
Female
Generation M F
1
2
3
4
Male
Female
Generation M F
1
2
3
4
Male
Female
Generation M F
1
2
3
4
Male
Female
Generally Similar to Men Across Time
Generation M F
1
2
3
4
Male
Female
Generally
Different from Men
Across Time
Chain Shift/Parallel Shift Components
Chain Shift Close or merged realization of BOT/BOUGHT Lowering and retraction of the nucleus of BAT Found among both post-World War II born middle class women
and men
Chain Shift/Parallel Shift Components
Chain Shift Close or merged realization of BOT/BOUGHT Lowering and retraction of the nucleus of BAT Found among both post-World War II born middle class
women and men
Parallel Shift Parallel retraction of BET and BIT, linked to the retraction of
BAT Found primarily among post-World War II born middle class
men
How Parallel Backing Works: Analogy
The Parallel Shift component appears to be advancing analogically in the vowel systems of speakers showing front vowel retraction tendencies
How Parallel Backing Works: Analogy
The Parallel Shift component appears to be advancing analogically in the vowel systems of speakers showing front vowel retraction tendencies
For BAT, post vocalic fricatives show stronger tendencies toward conditioning retracted realizations than post vocalic stops
fricatives
stops
fricatives
stops
fricatives
stops
fricatives
stops
fricatives
stops
Analogy and Parallel Backing
The Parallel Shift component appears to be advancing analogically in the vowel systems of speakers showing front vowel retraction tendencies
For BAT, post vocalic fricatives show stronger tendencies toward conditioning retracted realizations than post vocalic stops
This post vocalic conditioning division appears to be analogized from the BAT class to the BET class and to the BIT class by speakers on a vowel class by class basis.
Classical 4 part analogy? BAT -> (stops: fricatives) :: BET -> (stops: fricatives) BET -> (stops: fricatives) :: BIT -> (stops: fricatives)
Conclusions
In Columbus, the trajectory of BET and BIT appear to be more strongly backwards (at present) rather than backwards and diagonal (although data suggest this may occur in the future for BET)
The Canadian Shift in Columbus is comprised of two components:
Chain Shift involving close/merged realization of BOT and BOUGHT and the lowering and retraction of the nucleus of BAT
Parallel Shift of BET and BIT, linked to the retraction of the nucleus of BAT
The Parallel Shift component appears to be advancing analogically in the vowel systems of speakers showing front vowel retraction tendencies
Analogy and Parallel Movement
More generally, our findings suggest the mechanism of sound change for at least some vowel shifts is analogy (cf, Ohala, 1989, 1993; Hock, 2003; Janda & Joseph, 2003)
In particular, this appears to be true at least of parallel shifts:
A similar finding was previously discussed to some extent in regards to back vowel fronting (the back vowel parallel shift),* found in many dialects of English. (See Labov’s (1976) and Payne’s (1980) early studies of Philadelphia area speech for more details).
In our data here, we see a similar pattern for parallel movement again, albeit with front vowel retraction rather than back vowel fronting.
* That is, parallel fronting of the nuclei of /u:/, /oʊ/, and /aʊ/
Acknowledgements
For the loan of “old” data for use in the analysis: Bill Kretzschmar, on behalf of the Linguistic Atlas of the North Central States [LANCS]; Joan Hall, on behalf of the Dictionary of American Regional English [DARE]; Erik Thomas, for the use of a interview he recorded in 1984; and The Ohio Historical Society, for the use of 7 interviews recorded for various projects housed in their archives
For useful comments/advice: Brian Joseph, Cynthia Clopper, Don Winford, Mary Beckman, Erik Thomas, Hope Dawson, Bridget Smith, Na’im Tyson, Doug Bigham, and Facebook Friends
For assistance with vowel analysis: Jennifer Schumacher, Missy Reynard, and Emily Dorrian
…Coming Spring, 2010… A New Perspective on Vowel Variation across the 20th Century in
Columbus, OH by David Durian…