The Effect of Annotation Scheme Decisions on Parsing Learner Data

Marwa Ragheb and Markus Dickinson

Proceedings of the 13th International Workshop on Treebanks and Linguistic Theories (TLT13).

We present a study on the dependency parsing of second language learner data, focusing less on the parsing techniques and more on the effect of the linguistic distinctions made in the data. In particular, we examine syntactic annotation that relies more on morphological form than on meaning. We see the effect of particular linguistic decisions by: 1) converting and transforming a training corpus with a similar annotation scheme, with transformations occurring either before or after parsing; 2) inputting different kinds of part-of-speech (POS) information; and 3) analyzing the output. While we see a general favortism for parsing with more local dependency relations, this seems to be less the case for parsing the data of lower-level learners.


Electronically available file formats:


Bibtex entry:

@InProceedings{ragheb:dickinson:14a,
  author    = {Ragheb, Marwa and Dickinson, Markus},
  title     = {The Effect of Annotation Scheme Decisions on Parsing 
               Learner Data},
  booktitle = {Proceedings of the 13th International Workshop on 
               Treebanks and Linguistic Theories (TLT13},
  year      = {2014},
  address   = {T\"ubingen, Germany},
  pages     = {137--148},
  url       = {http://cl.indiana.edu/~md7/papers/ragheb-dickinson14a.html}
}