Korpus Pemelajar Bahasa Indonesia Beranotasi Eror (Koper): The First Indonesian Learner Corpus
David Moeljadi
Kanda University of International Studies
Korpus Pemelajar Bahasa Indonesia Beranotasi Eror (Koper):
The First Indonesian Learner Corpus
Learner corpora play a crucial role in second language acquisition research and
foreign language pedagogy, especially in investigating how a learner’s first language (L1)
influences the learning of a second language (L2). While several major languages such as
English, Japanese, and Chinese have well-established learner corpora—for instance, the
Michigan Corpus of Academic Spoken English (MICASE; Simpson et al., 1999) and the
International Corpus Network of Asian Learners of English (ICNALE; Ishikawa, 2013)—
no such corpus has been developed for Indonesian until now. This presentation introduces
Korpus Pemelajar Bahasa Indonesia Beranotasi Eror (Koper), the first annotated learner
corpus of Indonesian as a foreign language.
Koper is an error-tagged corpus of learner essays developed to support both
linguistic research and pedagogical practices in Indonesian language education (Moeljadi,
2023). It contains compositions written by students studying Indonesian at seven
universities in Japan. Each essay is manually annotated using a comprehensive set of 62
error tags, organized into four main categories: lexical, grammatical, spelling, and other
errors. Annotation is performed using the UAM Corpus Tool (O’Donnell, 2008) by a team
of four Indonesian lecturers based in Japan.
The development of Koper began in April 2023 and is ongoing. As of June 2025,
the corpus includes 1,427 essays written by 351 students in 2023, comprising a total of
177,705 words, along with over 1,420 additional essays written by more than 260 students
in 2024. In addition to error annotation, Koper incorporates rich metadata and undergoes
systematic preprocessing to ensure data quality and consistency.
Koper has already been utilized in research to identify high-frequency
vocabulary used by learners, revealing interesting overlaps with both general Indonesian
corpora and the BIPA (Bahasa Indonesia untuk Penutur Asing) conversation corpus
(Moeljadi, 2024). Current studies focus on learner error patterns and the influence of
Japanese as L1 on Indonesian acquisition. Future expansions will involve collecting data
from learners in other countries to broaden the scope and applicability of the corpus.
By providing a robust empirical foundation for the analysis of learner language,
Koper is expected to significantly enhance the development of Indonesian teaching
materials, curriculum design, and teaching methodologies for foreign learners