Language Bias in Accident Investigations

http://lup.lub.lu.se/luur/download?func=downloadFile&recordOId=2971193&fileOId=2971195

• The paper examined the language and content of the U.S. Forest Service’s Serios Accident Investigation Guide (SAIG)
• Idea: check if it biased the accident analysis process (tl;dr - yes)
• The Influence of Philosophy on Accident Investigation
  • The influence is large.
  • Today’s models are based on ‘cause-end-effect’ reasoning
  • They follow the need to create a logical pattern of ‘facts’ and ‘conclusions’
  • Even when the existence of a pattern is questionable
• The evolution of Accident Models
  • Sequential accident model
    • WWI -> accidents in a munition factory. Some people are accident prone -> remove
    • 1931, Heinrich developed first sequential accident model ‘Domino Theory’
    • => Humans were responsible for 80% of accidents, 20% were caused by unsafe conditions
    • It worked until the middle of 20 Century
    • Accident = uncontrolled energy release (physical impact)
  • Epidemiological Accident Model (Latent Failure Model)
    • 1978, Barry Turner developed ‘Man-Made Disaster’.
    • Accident - organizational and sociological phenomena.
    • It is not the direct result of recent events or human acts, rather - result of long process
    • 1987 (after Chernobyl), James Reason - “Epidemiological Accident Model)
    • “All man-made systems have within them the seeds of their own destruction, like ‘resident pathogen’ in the human body”
    • No direct responsible person
  • Systemic Accident Model
    • Accident - is a normal part of a complex system. Inevitable part.
    • Accident - is not a result of sequence of causal events or failures.
• Language as Communication and Structure
  • Locke, Chomsky and Pinker: mental thought occurs prior to the formation of language
  • Whorf, Sapir, Wittgenstein: knowledge is created through language
  • People use “cognitive economy” - a less detailed method of language to communicate
  • => Shared experience is used -> problem if it is not shared
  • People desire simple explanation -> binary opposition.
  • => ‘either-or’ thinking can limit possibilities
  • The assignment of agency to an event is highly influenced by the language that is spoken
  • i.e. English speakers - stronger tendency to assign agency to even unintentional events
  • e.g. “active verb voice” -> more likely to blame someone
• Findings in the SAIG instructions …
• My Summary
  • Active voice -> sense of control and causation that is lacking in the passive voice
  • Cognitive economy may fail to recognize more meanings in input data, especially if no shared background
  • No proper shared definition -> unintended conclusions
  • People by default uses Blame Culture
  • Sequential accident model is widely known, while others are not -> causality, human responsibility for everything