---

Corrigendum

---

Concerning: Willem Zuidema, Theoretical Evaluation of Estimation Methods for Data-Oriented Parsing, in: Conference Companion / Proceedings 11th Conference of the European Chapter of the Association for Computational Linguistics (EACL'06), published by the Association for Computational Linguistics, pp. 183-186

There is an unfortunate error in the published version of this paper, in the way I represented the paper of Mark Johnson (2002). On p.183 I wrote that the "DOP1 method really is biased and inconsistent, but not for the reasons Johnson gives", which is incorrect, because Johnson's paper in fact contains a complete proof of the bias and inconsistency of DOP1. On p.185 I give an example that shows this again (eq.11), but here I should have acknowledged Johnson's proof. My argument for what I called the "frequency-distribution test" on p. 184 still holds, but Johnson made the same point much earlier when he wrote (p.73):

"it is more natural to define bias and loss in terms of the probability distributions that the parameters specify, rather than in terms of the parameters themselves."

Although the suggested contrast was never meant to be the main point of my paper, I regret to have been inaccurate in my reference to earlier work, and apologize to Mark Johnson and anyone that might have been confused and annoyed by the error. The version of this paper on my website has been corrected.

Willem Zuidema

Reference:

Mark Johnson (2002). The DOP estimation method is biased and inconsistent. Computational Linguistics, 28(1):71-76.