Introduction

This page describes the annotation of a Turkish “GB” treebank, a treebank of grammar-book examples annotated according to Universal Dependencies (UD) annotation scheme. The treebank aims to cover a large range of grammatical constructions in the language with realtively small number of short sentences or sentence fragments. It also has served as the first attempt to adapt UD annotation scheme to Turkish, and the present UD Turkish guidelines are based on the documentation of this treebank. The treebank was introduced in Çöltekin (2015), contains sentences from Göksel and Kerslake (2005) and addeitions from Kornfilt 1997 are on the way).

As of version 2.4, the treebank contains 2802 sentences, and 16509 tokens.

Sentence selection, preprocessing, toekenization

The treebank contains all numbered examples from the grammar book. A few in-text examples are also included. Some of the examples (currently 409) are not sentences but sentence fragments, typically phrases. The full sentences in the treebank always end with a punctuation mark, while sentence fragments are are without a final punctuation. The ungrammatical examples are skipped. The “borderline acceptable” examples (marked with ? or ??) were included, with a sentence-level comment # acceptability = ? (the value is as indicated in the grammar book).

References