Manipulative Language Strategies in Tobacco Industry Media Statements

1
Manipulative Language Strategies in Tobacco
Industry Media Statements

• Donald Rubin
• Norbert Hirschhorn
• Lona Jean Turner

2
Linguistic Analyses of Tobacco Industry Documents
NIH/NCI RO1 CA87490 -conclusions are solely the
views of the authors
3
Two Kinds of Document Samples

• Rhetorical Samples
• Multiple draft cases
• Cross-audience cases
• Internal audience ? in employ of tobacco
• External audience? evidence of dissemination
  (e.g., press statements)
  
• Linked by topic, date

• Stratified Random Sample (Quota)
• Random selection within decade, /- internal
  audience, /- named audience
  
• Quotas established by earlier core sample
• 500,000 words (800 documents)
• Functions as baseline for comparing biased
  samples

4
Primary On-Line Sites for Searching Tobacco
Industry Documents

• Legacy Library http//legacy.library.ucsf.edu/
• Tobacco Documents On-line
• http//tobaccodocuments.org/
• Industry Sites http//www.tobaccoarchives.com/
• E.g., http//www.pmdocs.com/

5
Samples for Present Study

• Press Statements
• Evidence of dissemination (e.g., letterhead)
• N11,212 words 18 documents
• Internal Counterpart Documents
• Includes in-house memos, position development
• N14213 words 18 documents
• Quota Sample
• N268338 words 400 documents
• Not participating in statistical testing

6
Computer-Assisted Language Analysis XML Coding
7
Computer-Assisted Language Analysis Concordance
Comparison

• Relative frequency of each type (word) divided by
  total tokens.
  
• Z-test for equal proportions (?.05)
• Allows comparison between any two documents or
  between sets of documents (e.g., external
  audience docs vs. reference corpus)
  
• Authored by Clayton Darwin, UGA

8
Computer-Assisted Language Analysis DICTION

• Contains 32 dictionaries of rhetorical themes
  (e.g., temporal, denial, blame, aggression)
  
• Yields 5 composite dimensions certainty,
  activity, commonality, realism, optimism
  
• Yields 4 stylistic properties variety,
  embellishment, insistence, complexity
  
• Provides norms for variety of genres (e.g.,
  corporate relations, based on 163 specimens)
  
• Authored by Rod Hart, UTexas

9
Computer-Assisted Language Analysis D. Bibers
Corpus Linguistics
10
Findings Press statements use fewer hedges, more
certain language than internal counterpart
documents
11
Findings Press statements invoke popular themes,
avoid mention of disease
12
Findings Internal counterpart docs create
identity with insiders Press statements
establish a 3rd-person stance
13
Findings Press statements repeat key words
excessively other stylistics are normal for PR
genre
14
Findings Internal counterparts, press statements
are within most genre norms
15
Findings Press statements are low on praise,
satisfaction high on certainty
16
General Conclusions

• Extensive on-line (and depository) searching is
  necessary to identify internal counterparts to
  publicly released tobacco industry statements
  
• Computer-assisted linguistic analysis is a
  scientific means for establishing differential
  language use in external vs internal documents
  
• Tobacco industry press statements are
  characterized by repetition of key words, use of
  popular themes, 3rd person stance, and avoidance
  of disease terms, high certainty, and high blame
  and negativism (low optimism)