Essay:Quantifying Liberal Style

This is an old revision of this page, as edited by MarkGall (talk | contribs) at 01:26, October 7, 2009. It may differ significantly from current revision.

Quantifying Liberal Style

Statistical methods have been used to quantify a wide range of characteristics of writing. For example, the Fleisch-Kincaid grade level and reading ease scores may be computed by almost any word processor, including many versions of Microsoft Word. These scores attempt to assess the level of sophistication of samples of writing, and have been used to give concrete evidence of a dumbing-down of presidential inaugural addresses over time. Great Britain is even considering implementing computerized grading of essays on some standardized tests. Given the objective nature of many of the characteristics of liberal style identified at Conservapedia, it is natural to devise analogous measures of the degree of liberal style of a specimen of writing.

A rough formula to measure the liberal style of a passage, adapted to Wiki-style websites, is

Failed to parse (MathML with SVG or PNG fallback (recommended for modern browsers and accessibility tools): Invalid response ("Math extension cannot connect to Restbase.") from server "https://wikimedia.org/api/rest_v1/":): {\displaystyle \textrm{Liberality} = C_0 \cdot \frac{\textrm{TalkPst}}{\textrm{MainPst}} + C_1 \cdot \frac{\textrm{TalkWd}}{\textrm{TalkPst}} + C_2 \cdot \left( \lambda \cdot \textrm{LibWd} - \textrm{ConWd} \right) + C_3 \,\textrm{LibPhr}^2 + C_4 \, \textrm{LastWd} } ,

where:

  • MainPst = number of posts to mainspace pages.
  • TalkPst = number of posts to talk pages.
  • TalkWd = number of words posted to talk pages.
  • LibWd = number of uses of liberal words from reference list
  • ConWd = number of uses of conservative words from reference list
  • Failed to parse (MathML with SVG or PNG fallback (recommended for modern browsers and accessibility tools): Invalid response ("Math extension cannot connect to Restbase.") from server "https://wikimedia.org/api/rest_v1/":): {\displaystyle \lambda} = (number of conservative words in reference list)/(number of liberal words in reference list).
  • LibPhr = number of uses of liberal favorite phrases identified in Liberal style, such as "silence speaks volumes" and "reflects poorly on the site"
  • LastWd = number of instances of last wordism in user's talk page posts.

Additionally, the Failed to parse (MathML with SVG or PNG fallback (recommended for modern browsers and accessibility tools): Invalid response ("Math extension cannot connect to Restbase.") from server "https://wikimedia.org/api/rest_v1/":): {\displaystyle C_i} are constants which must be determined through careful testing. For now, we assume that C0 = 3, C1 = 1, C2 = 200, C3 = 400, C4 = 50. Assuming that all talk page posts by all editors have the same number of ideas on average (but not the same length), the liberality index given above has units words/idea, roughly corresponding to the trait of liberal longwindedness (here the constants must be understood with appropriate units).

The quantification of liberal style has an astonishing array of possible applications. Most directly, it may be applied to the automatic detection of vandals on Conservapedia who attempt to insert liberal bias undetected. Related indices of liberal bias (not based on a Wiki-style count) could also be used to measure the degree to which various Bible translations have been compromised by liberal bias, and for quality control on books in the Conservative Bible Project. Indeed, linguistic indicators such as vocabulary use have long been used by scholars to gauge he authenticity of various ancient texts (including the Bible), and applying a similar metric (adapted to other languages) to the original text of the Bible may help discover additional passages which are late additions. At last, measures of liberal style may be applied to measure the political bias of various news outlets, and of specific anchors at those outlets.

While some have questioned the usefulness of quantifying things like open-mindedness and liberal style (see point 30 on Liberal style), the success of similar metrics for other aspects of language indicates that such calculations are indeed very meaningful.