Proceedings of the Sixth Linguistic Annotation Workshop, Jul 12, 2012
We present a novel scheme for annotating the realization and ellipsis of Korean particles. Annota... more We present a novel scheme for annotating the realization and ellipsis of Korean particles. Annotated data include 100,128 Ecel (a spacebased word unit) in spoken and written corpora composed of four different genres in order to evaluate how register variation contributes to Korean particle ellipsis. Identifying the grammatical functions of particles and zero particles is critical for deriving a valid linguistic analysis of argument realization, semantic and discourse analysis, and computational processes of parsing. The primary challenge is to design a reliable scheme for classifying particles while making a clear distinction between ellipsis and non-occurrences. We determine in detail issues involving particle annotation and present solutions. In addition to providing a statistical analysis and outcomes, we briefly discuss linguistic factors involving particle ellipsis.
Proceedings of the Fourth Linguistic Annotation Workshop, Jul 15, 2010
This paper presents preliminary work on a corpus-based study of Korean demonstratives. Through th... more This paper presents preliminary work on a corpus-based study of Korean demonstratives. Through the development of an annotation scheme and the use of spoken and written corpora, we aim to determine different functions of demonstratives and to examine their distributional properties. Our corpus study adopts similar features of annotation used in Botley and McEnery (2001) and provides some linguistic hypotheses on grammatical functions of Korean demonstratives to be further explored.
We further work on detecting errors in postpositional particle usage by learners of Korean by imp... more We further work on detecting errors in postpositional particle usage by learners of Korean by improving the training data and developing a complete pipeline of particle selection. We improve the data by filtering non-Korean data and sampling instances to better match the particle distribution. Our evaluation shows that, while the data selection is effective, there is much work to be done with preprocessing and system optimization.
Proceedings of the International Conference on Head-Driven Phrase Structure Grammar
Some examples of non-object TCs are given as follows. i Lazarusj-ka [ j shyophingha-ki ]-ka/ey sw... more Some examples of non-object TCs are given as follows. i Lazarusj-ka [ j shyophingha-ki ]-ka/ey swipta (Locative) Lazarus-nom do shopping-nml-nom/for easy 'Lazarus is easy to do shopping (in)' ii yenphilj-i [ j kulssi-lul sseu-ki ]-ka/ey himtulta (Instrumental) a pencil-nom letters-acc write-nml-nom/for tough 'A pencil is tough to write letters (with).'
We aim to sufficiently define annotation for post-positional particle errors in L2 Korean writing... more We aim to sufficiently define annotation for post-positional particle errors in L2 Korean writing, so that future work on automatic particle error detection can make progress. To achieve this goal, we outline the linguistic properties of Korean particles in learner data. Given the agglutinative nature of Korean and the range of functions of particles, this annotation effort involves issues such as defining the tokens and target forms.
We present a novel scheme for annotating the realization and ellipsis of Korean particles. Annota... more We present a novel scheme for annotating the realization and ellipsis of Korean particles. Annotated data include 100,128 Ecel (a space-based word unit) in spoken and written corpora composed of four different genres in order to evaluate how register variation contributes to Korean particle ellipsis. Identifying the grammatical functions of particles and zero particles is critical for deriving a valid linguistic analysis of argument realization, semantic and discourse analysis, and computational processes of parsing. The primary challenge is to design a reliable scheme for classifying particles while making a clear distinction between ellipsis and non-occurrences. We determine in detail issues involving particle annotation and present solutions. In addition to providing a statistical analysis and outcomes, we briefly discuss linguistic factors involving particle ellipsis.
Post-positional particles are a significant source of errors for learners of Korean. Following me... more Post-positional particles are a significant source of errors for learners of Korean. Following methodology that has proven effective in handling English preposition errors, we are beginning the process of building a machine learner for particle error detection in L2 Korean writing. As a first step, however, we must acquire data, and thus we present a methodology for constructing large-scale corpora of Korean from the Web, exploring the feasibility of building corpora appropriate for a given topic and grammatical construction.
We detect errors in Korean post-positional particle usage, focusing on optimizing omission detect... more We detect errors in Korean post-positional particle usage, focusing on optimizing omission detection, as omissions are the single-biggest factor in particle errors for learners of Korean. We also develop a system for predicting the correct choice of a particle. For omission detection, we model the task largely on English grammatical error detection, but employ Korean-specific features and filters; likewise, output analysis and the omission correction system illustrate how unique properties of Korean, such as the distinct types of particles used, need to be accounted for in adapting the system, thereby moving the field one step closer to robust multi-lingual methods.
This paper presents preliminary work on a corpus-based study of Korean demonstratives. Through th... more This paper presents preliminary work on a corpus-based study of Korean demonstratives. Through the development of an annotation scheme and the use of spoken and written corpora, we aim to determine different functions of demonstratives and to examine their distributional properties. Our corpus study adopts similar features of annotation used in Botley and McEnery (2001) and provides some linguistic hypotheses on grammatical functions of Korean demonstratives to be further explored.
This paper explores several important issues in developing syntactically annotated Korean corpora... more This paper explores several important issues in developing syntactically annotated Korean corpora for higher-level language processing, including semantic-discourse parsing, question-answering, machine translation, information retrieval, etc. In particular, we compare the Penn Korean Treebank (PKT) and the Korean Treebank of the 21st Century Sejong Project (ST) and discuss four critical issues in syntactic annotation. We argue for the use of more sophisticated morphosyntactic information, and based on our comparative study, we propose revisions in the syntactic annotation schemes of the existing Korean Treebanks in order to improve the quality of annotated corpora and their usability both for conducting theoretical research and for developing computational tools. The results of our comparative study reveal four significant issues in syntactic annotations: the syntactic analysis of verbal complexes, the hierarchical structure of noun phrases, the representation of traces, and the mar...
This paper argues for the necessity of zero pronoun annotations in Korean treebanks and provides ... more This paper argues for the necessity of zero pronoun annotations in Korean treebanks and provides an annotation scheme that can be used to develop a gold standard for testing different anaphor resolution algorithms. Relevant issues of pronoun annotation will be discussed by comparing the Penn Korean Treebank with zero pronoun markup and the newly developing Sejong Teebank without zero pronoun markup. In addition to supportive evidence for zero marking, necessary morphosyntactic and semantic features will be suggested for zero annotation in Korean treebanks.
This paper argues for the necessity of zero pronoun annotations in Korean treebanks and provides ... more This paper argues for the necessity of zero pronoun annotations in Korean treebanks and provides an annotation scheme that can be used to develop a gold standard for testing different anaphor resolution algorithms. Relevant issues of pronoun annotation will be discussed by comparing the Penn Korean Treebank with zero pronoun markup and the newly developing Sejong Teebank without zero pronoun markup. In addition to supportive evidence for zero marking, necessary morphosyntactic and semantic features will be suggested for zero annotation in Korean treebanks.
Some examples of non-object TCs are given as follows. i Lazarusj-ka [ j shyophingha-ki ]-ka/ey sw... more Some examples of non-object TCs are given as follows. i Lazarusj-ka [ j shyophingha-ki ]-ka/ey swipta (Locative) Lazarus-nom do shopping-nml-nom/for easy 'Lazarus is easy to do shopping (in)' ii yenphilj-i [ j kulssi-lul sseu-ki ]-ka/ey himtulta (Instrumental) a pencil-nom letters-acc write-nml-nom/for tough 'A pencil is tough to write letters (with).'
In this study, we focus on particle errors and discuss an annotation scheme for Korean learner co... more In this study, we focus on particle errors and discuss an annotation scheme for Korean learner corpora that can be used to extract heuristic patterns of particle errors efficiently. We investigate different properties of particle errors so that they can be later used to identify learner errors automatically, and we provide resourceful annotation guidelines. We present issues that are relevant to learner error annotation including how to classify particle error types, present correct tokens, mark overlapping error types, and so on. Accurate annotation of particle errors will be useful for extracting relevant error rules and will provide substantial benefits to the development of ICALL systems. We argue that it is necessary to link annotation with feedback procedures to increase the efficiency of systems. Furthermore, we investigate whether heritage learners and nonheritage learners generate different error patterns and discuss significant implications for heritage versus nonheritage language learning.
Proceedings of the Sixth Linguistic Annotation Workshop, Jul 12, 2012
We present a novel scheme for annotating the realization and ellipsis of Korean particles. Annota... more We present a novel scheme for annotating the realization and ellipsis of Korean particles. Annotated data include 100,128 Ecel (a spacebased word unit) in spoken and written corpora composed of four different genres in order to evaluate how register variation contributes to Korean particle ellipsis. Identifying the grammatical functions of particles and zero particles is critical for deriving a valid linguistic analysis of argument realization, semantic and discourse analysis, and computational processes of parsing. The primary challenge is to design a reliable scheme for classifying particles while making a clear distinction between ellipsis and non-occurrences. We determine in detail issues involving particle annotation and present solutions. In addition to providing a statistical analysis and outcomes, we briefly discuss linguistic factors involving particle ellipsis.
Proceedings of the Fourth Linguistic Annotation Workshop, Jul 15, 2010
This paper presents preliminary work on a corpus-based study of Korean demonstratives. Through th... more This paper presents preliminary work on a corpus-based study of Korean demonstratives. Through the development of an annotation scheme and the use of spoken and written corpora, we aim to determine different functions of demonstratives and to examine their distributional properties. Our corpus study adopts similar features of annotation used in Botley and McEnery (2001) and provides some linguistic hypotheses on grammatical functions of Korean demonstratives to be further explored.
We further work on detecting errors in postpositional particle usage by learners of Korean by imp... more We further work on detecting errors in postpositional particle usage by learners of Korean by improving the training data and developing a complete pipeline of particle selection. We improve the data by filtering non-Korean data and sampling instances to better match the particle distribution. Our evaluation shows that, while the data selection is effective, there is much work to be done with preprocessing and system optimization.
Proceedings of the International Conference on Head-Driven Phrase Structure Grammar
Some examples of non-object TCs are given as follows. i Lazarusj-ka [ j shyophingha-ki ]-ka/ey sw... more Some examples of non-object TCs are given as follows. i Lazarusj-ka [ j shyophingha-ki ]-ka/ey swipta (Locative) Lazarus-nom do shopping-nml-nom/for easy 'Lazarus is easy to do shopping (in)' ii yenphilj-i [ j kulssi-lul sseu-ki ]-ka/ey himtulta (Instrumental) a pencil-nom letters-acc write-nml-nom/for tough 'A pencil is tough to write letters (with).'
We aim to sufficiently define annotation for post-positional particle errors in L2 Korean writing... more We aim to sufficiently define annotation for post-positional particle errors in L2 Korean writing, so that future work on automatic particle error detection can make progress. To achieve this goal, we outline the linguistic properties of Korean particles in learner data. Given the agglutinative nature of Korean and the range of functions of particles, this annotation effort involves issues such as defining the tokens and target forms.
We present a novel scheme for annotating the realization and ellipsis of Korean particles. Annota... more We present a novel scheme for annotating the realization and ellipsis of Korean particles. Annotated data include 100,128 Ecel (a space-based word unit) in spoken and written corpora composed of four different genres in order to evaluate how register variation contributes to Korean particle ellipsis. Identifying the grammatical functions of particles and zero particles is critical for deriving a valid linguistic analysis of argument realization, semantic and discourse analysis, and computational processes of parsing. The primary challenge is to design a reliable scheme for classifying particles while making a clear distinction between ellipsis and non-occurrences. We determine in detail issues involving particle annotation and present solutions. In addition to providing a statistical analysis and outcomes, we briefly discuss linguistic factors involving particle ellipsis.
Post-positional particles are a significant source of errors for learners of Korean. Following me... more Post-positional particles are a significant source of errors for learners of Korean. Following methodology that has proven effective in handling English preposition errors, we are beginning the process of building a machine learner for particle error detection in L2 Korean writing. As a first step, however, we must acquire data, and thus we present a methodology for constructing large-scale corpora of Korean from the Web, exploring the feasibility of building corpora appropriate for a given topic and grammatical construction.
We detect errors in Korean post-positional particle usage, focusing on optimizing omission detect... more We detect errors in Korean post-positional particle usage, focusing on optimizing omission detection, as omissions are the single-biggest factor in particle errors for learners of Korean. We also develop a system for predicting the correct choice of a particle. For omission detection, we model the task largely on English grammatical error detection, but employ Korean-specific features and filters; likewise, output analysis and the omission correction system illustrate how unique properties of Korean, such as the distinct types of particles used, need to be accounted for in adapting the system, thereby moving the field one step closer to robust multi-lingual methods.
This paper presents preliminary work on a corpus-based study of Korean demonstratives. Through th... more This paper presents preliminary work on a corpus-based study of Korean demonstratives. Through the development of an annotation scheme and the use of spoken and written corpora, we aim to determine different functions of demonstratives and to examine their distributional properties. Our corpus study adopts similar features of annotation used in Botley and McEnery (2001) and provides some linguistic hypotheses on grammatical functions of Korean demonstratives to be further explored.
This paper explores several important issues in developing syntactically annotated Korean corpora... more This paper explores several important issues in developing syntactically annotated Korean corpora for higher-level language processing, including semantic-discourse parsing, question-answering, machine translation, information retrieval, etc. In particular, we compare the Penn Korean Treebank (PKT) and the Korean Treebank of the 21st Century Sejong Project (ST) and discuss four critical issues in syntactic annotation. We argue for the use of more sophisticated morphosyntactic information, and based on our comparative study, we propose revisions in the syntactic annotation schemes of the existing Korean Treebanks in order to improve the quality of annotated corpora and their usability both for conducting theoretical research and for developing computational tools. The results of our comparative study reveal four significant issues in syntactic annotations: the syntactic analysis of verbal complexes, the hierarchical structure of noun phrases, the representation of traces, and the mar...
This paper argues for the necessity of zero pronoun annotations in Korean treebanks and provides ... more This paper argues for the necessity of zero pronoun annotations in Korean treebanks and provides an annotation scheme that can be used to develop a gold standard for testing different anaphor resolution algorithms. Relevant issues of pronoun annotation will be discussed by comparing the Penn Korean Treebank with zero pronoun markup and the newly developing Sejong Teebank without zero pronoun markup. In addition to supportive evidence for zero marking, necessary morphosyntactic and semantic features will be suggested for zero annotation in Korean treebanks.
This paper argues for the necessity of zero pronoun annotations in Korean treebanks and provides ... more This paper argues for the necessity of zero pronoun annotations in Korean treebanks and provides an annotation scheme that can be used to develop a gold standard for testing different anaphor resolution algorithms. Relevant issues of pronoun annotation will be discussed by comparing the Penn Korean Treebank with zero pronoun markup and the newly developing Sejong Teebank without zero pronoun markup. In addition to supportive evidence for zero marking, necessary morphosyntactic and semantic features will be suggested for zero annotation in Korean treebanks.
Some examples of non-object TCs are given as follows. i Lazarusj-ka [ j shyophingha-ki ]-ka/ey sw... more Some examples of non-object TCs are given as follows. i Lazarusj-ka [ j shyophingha-ki ]-ka/ey swipta (Locative) Lazarus-nom do shopping-nml-nom/for easy 'Lazarus is easy to do shopping (in)' ii yenphilj-i [ j kulssi-lul sseu-ki ]-ka/ey himtulta (Instrumental) a pencil-nom letters-acc write-nml-nom/for tough 'A pencil is tough to write letters (with).'
In this study, we focus on particle errors and discuss an annotation scheme for Korean learner co... more In this study, we focus on particle errors and discuss an annotation scheme for Korean learner corpora that can be used to extract heuristic patterns of particle errors efficiently. We investigate different properties of particle errors so that they can be later used to identify learner errors automatically, and we provide resourceful annotation guidelines. We present issues that are relevant to learner error annotation including how to classify particle error types, present correct tokens, mark overlapping error types, and so on. Accurate annotation of particle errors will be useful for extracting relevant error rules and will provide substantial benefits to the development of ICALL systems. We argue that it is necessary to link annotation with feedback procedures to increase the efficiency of systems. Furthermore, we investigate whether heritage learners and nonheritage learners generate different error patterns and discuss significant implications for heritage versus nonheritage language learning.
Uploads
Papers by Sun-Hee Lee