West Coast Contractions

1
West Coast Contractions

• Jack Grieve
• Northern Arizona University
• AAACL 2006

2
Research Questions

• What is the extent of regional variation in
  natural language?
• Does a correlation exist between contraction
  rates in letters to the editor written by authors
  from the American West Coast and the latitude of
  the authors places of residence?

3
The Variety

• Register
• Published Newspaper Letters to the Editor
• Era
• 2005
• Region
• West Coast, United States

4
Register

• Published Letters to the Editor
• Writing less likely than speech to contain
  regional variation.
• Published English less likely than unpublished
  English to contain regional variation.
• Letters to the Editor are written by many
  non-professional authors, from many regions.
• Letters to the Editor are freely and abundantly
  available online in machine readable form.

5
Region

• West Coast, United States
• Little regional variation has been found within
  this region in previous dialect studies (Reed,
  1954 Reed, 1956, 1957 Kurath, 1972 Carver
  1987 Wolfram and Schilling-Estes, 1998 Labov,
  Ash and Boberg, 2006).
• Recently settled region.
• Well- and evenly-settled region.
• Geographically well-defined, allowing for
  regional variation to be measured as a
  correlation between feature measure and latitude.

6
(No Transcript)
7
Corpus Compilation

• Manually download letters from 13 online
  newspaper archives (approx. 1,000 letters and
  200,000 words per newspaper).
• Clean up letters and sort by place of residence
  of author.
• Form a sub-corpus for each city with a least 50
  letters to the editor and 50 contraction
  environments.
• Tag corpora (using the Biber Tagger).

8
Corpus Dimensions

• Total city sub-corpora 28
• Bellingham, Seattle, Ferndale, Bellingham,
  Kirkland, Redmond, Bellevue, Seattle, Olympia,
  Lacey, Portland, Keizer, Salem, Eugene,
  Springfield, Medford, Chico, Fair Oaks,
  Roseville, Carmichael, Sacramento, Davis, Elk
  Grove, Berkeley, Oakland, San Francisco, Clovis,
  Fresno, Bakersfield, La Jolla and San Diego.
• Total words 2,084,968
• Total letters 10,767
• Total Contraction Environments 59,691

9
(No Transcript)
10
Contractions

• Frequent in Written English.
• Variable in Written English.
• Easy to count mechanically.
• Not edited in Letters to the Editor.
• One of the most often analyzed grammatical
  features in sociolinguistic studies (e.g. Kurath,
  1939 Labov, 1969 Wolfram, 1969, 1974 Baugh,
  1980 Biber, 1987 McElhinny, 1993 Rickford,
  1998 Yaeger-Dror, Hall-Lew, Deckert, 2002).

11
(Verb-)Not-Contraction

• Modal They cant improve a stadium. (Salem)
• Passive I wasnt asked. (Lacey)
• Prog. Schools arent offering healthy
  items. (Chico)
• Perfect If you havent looked yet, too bad.
  (Bellingham)
• Aux Do They don't want to make their lives
  better.
• (Bakersfield)
• Copula These arent gifts or grants. (Portland)

12
(Pronoun-)Verb-Contraction

• Modal They'll just find another Anita
  Hill. (Bakersfield)
• Passive Its too blinded by greed. (San
  Diego)
• Prog. Theyre trying to escape the
  pain. (Seattle)
• Perfect Ive lived here 30 years without
  issue. (Chico)
• Copula Its not electioneering, I
  guess. (Eugene)

13
Contraction Rate

• Not- and Verb-contraction rates calculated for
  each of the 28 city sub-corpora.
• Contraction Rate
• Total Number of Contractions
• Total Number of Contraction Environments

14
Data Analysis

• Test for linear correlation (Pearsons
  Correlation Coefficient) between not-contraction
  rates and decimal latitude and verb-contraction
  rates and decimal latitude across the 28 city
  sub-corpora.

15
Not-contraction
r 0.640
16
Verb-contraction
r 0.584
17
Formality

• Contraction is a feature associated with informal
  language, therefore the pattern observed may be
  part of a larger patternNortherners on the West
  Coast writing more informally than Southerners.
• To test this hypothesis, the latitude correlation
  tests are repeated with other features associated
  with informality (Biber, 1988).

18
Results
19
Conclusion

• Statistically significant regional patterns found
  in West Coast letters to the editora variety of
  language where no such variation would be
  expected to be found.
• This is the first time that regional variation
  has been found in writing at the sub-national
  level.
• Contraction pattern appears to be part of a
  larger formality pattern.
• It is unclear why this formality pattern exists.