Gheg Albanian Pear Stories (GPS) contain renarrations of Wallace Chafe's Pear Stories video (pearstories.org) by heritage speakers of Gheg Albanian living in Switzerland and speakers from Prishtina.
The corpus is currently available as a Universal Dependency (UD) treebank at the repository [UniversalDependencies/UD_Gheg-GPS](https://github.com/UniversalDependencies/UD_Gheg-GPS)
# Introduction
UD Gheg GPS contains 966 sentences from 64 recordings of Gheg speakers re-narrating the *Pear Stories* video. Data collection was part of a bigger project that took place from May 2019 to July 2022 in Zurich, Prishtina and Munich. Only recordings from Prishtina und Zurich were included in the treebank. Speakers of three different generations were interviewed, age ranging from 10 to 67. Sentence ids contain information on location (`P` for Prishtina, `Z` for Zurich), Generation (`G1`, `G2`, `G3`) and a unique speaker id, all separated by hyphens, followed by an underscore and the sentence id, which starts at 1 for each interview. Due to the multilingual setting, the treebank contains many instances of code-switching (mostly Swiss-German). It also exhibits characteristics of (semi-)spontaneous speech, like disfluencies and corrections.
The treebank contains 16k tokens and was not split into training and test set.
