Corpora: T-score

[Tom Emerson]

Ha Le An writes:
 > Sorry to write this letter, but after a lot of attempts, I can not find a
 > clear definition for T-score. Can anybody help me?

Any of the following references are excellent sources on this:

@Book{Oakes:1998:SCL,
  author =       {Oakes, Michael P.},
  title =        {Statistics for Corpus Linguistics},
  publisher =    {Edinburgh University Press},
  year =         {1998},
  isbn =         {0748608176}
}

@Book{Manning:1999:FSN,
  author =       {Christopher D. Manning and Hinrich Sch\"utze},
  title =        {Foundations of Statistical Natural Language Processing},
  publisher =    MIT,
  year =         {1999},
  isbn =         {0262133601}
}

@Book{Charniak:1993:SLL,
  author =       {Eugene Charniak},
  title =        {Statistical Language Learning},
  publisher =    MIT,
  year =         {1993},
  isbn =         {0262531410}
}

To briefly quote from Charniak, p. 122:

"The t-score is a standard is a standard test to see if the results
could be expected on the basis of chance."

      -tree

--
Tom Emerson                                          Basis Technology Corp.
Language Hacker                                    http://www.basistech.com
  "Beware the lollipop of mediocrity: lick it once and you suck forever"