• Speech Sciences
• /
• v.15 no.1
• /
• pp.7-23
• /
• 2008

This paper investigates tonal patterns of the prosodic constituents of an AP and a PWD in Korean and their relation with the morphological/syntactic structure. Specifically, this paper asks the following questions: First, if there are more than one PWD in an AP, how is each PWD specified in terms of tones? Secondly, in case that there is only one PWD in an AP that consists of several morphemes, is there any preference of the association between tones and the morphemes that constitute that PWD? Thirdly, if an AP dominates a PWD and if a PWD contains at least one morpheme of the lexical category, it follows that an AP should contain at least one morpheme of the lexical category. Can this be verified with the experimental data? In order to answer these questions, Experiment I and II were conducted with the target material consisting of a stem and suffixes that varied in length. The results of this preliminary test show that as the number of syllables in the target material increases, the more number of an AP tonal pattern occurs in it and as a result, in some cases, an AP consisting of suffixes only may occur.

• Journal of KIISE
• /
• v.44 no.1
• /
• pp.36-43
• /
• 2017

Syntactic information represents the dependency relation between predicates and arguments, and it is helpful for improving the performance of Semantic Role Labeling systems. However, syntax analysis can cause computational overhead and inherit incorrect syntactic information. To solve this problem, we exclude syntactic information and use only morpheme information to construct Semantic Role Labeling systems. In this study, we propose an end-to-end SRL system that only uses morpheme information with Stacked Bidirectional LSTM-CRFs model by extending the LSTM RNN that is suitable for sequence labeling problem. Our experimental results show that our proposed model has better performance, as compare to other models.

• Journal of KIISE
• /
• v.42 no.6
• /
• pp.817-822
• /
• 2015

Relation extraction is an important information extraction technique that can be widely used in areas such as question-answering and knowledge population. Previous studies on relation extraction have been based on supervised machine learning models that need a large amount of training data manually annotated with relation categories. Recently, to reduce the manual annotation efforts for constructing training data, distant supervision methods have been proposed. However, these methods suffer from a drawback: it is difficult to use these methods for collecting negative training data that are necessary for resolving classification problems. To overcome this drawback, we propose a one-class classification model that can be trained without using negative data. The proposed model determines whether an input data item is included in an inner category by using a similarity measure based on lexical information and syntactic patterns in a vector space. In the experiments conducted in this study, the proposed model showed higher performance (an F1-score of 0.6509 and an accuracy of 0.6833) than a representative one-class classification model, one-class SVM(Support Vector Machine).

• Journal of the Korea Society of Computer and Information
• /
• v.27 no.3
• /
• pp.13-23
• /
• 2022

In this study we leveraged a probing method to evaluate how a pre-trained L2 LSTM language model represents sentences with relative and coordinate clauses. The probing experiment employed adapted models based on the pre-trained L2 language models to trace the syntactic properties of sentence embedding vector representations. The dataset for probing was automatically generated using several templates related to different sentence structures. To classify the syntactic properties of sentences for each probing task, we measured the adaptation effects of the language models using syntactic priming. We performed linear mixed-effects model analyses to analyze the relation between adaptation effects in a complex statistical manner and reveal how the L2 language models represent syntactic features for English sentences. When the L2 language models were compared with the baseline L1 Gulordava language models, the analogous results were found for each probing task. In addition, it was confirmed that the L2 language models contain syntactic features of relative and coordinate clauses hierarchically in the sentence embedding representations.

• Proceedings of the Korean Society for Language and Information Conference
• /
• 2007.11a
• /
• pp.415-421
• /
• 2007

This paper proposes a convolution tree kernel-based approach for relation extraction where the parse tree is expanded with entity features such as entity type, subtype, and mention level etc. Our study indicates that not only can our method effectively capture both syntactic structure and entity information of relation instances, but also can avoid the difficulty with tuning the parameters in composite kernels. We also demonstrate that predicate verb information can be used to further improve the performance, though its enhancement is limited. Evaluation on the ACE2004 benchmark corpus shows that our system slightly outperforms both the previous best-reported feature-based and kernel-based systems.

• Speech Sciences
• /
• v.12 no.1
• /
• pp.15-25
• /
• 2005

The purpose of the present study was to investigate the characteristics of metalinguistic awareness of children with specific language impairment. Forty-five children participated in this study; 15 children with specific language impairment (SLI group) whose range of the language age was 4;6-6;6, 15 normal children chronological age matched (CA controls) and 15 normal children language age matched with the SLI group (CA controls). A metalinguistic task involving the identification and revision of syntactic, semantic, and phonologic errors was used. The SLI group performed significantly poorer than CA controls as well as LA controls in identifying and correcting error sentences, especially sentences with syntactic error. These results revealed the relation between language problems of children with SLI and metalinguistic abilities.

• Language and Information
• /
• v.3 no.1
• /
• pp.1-20
• /
• 1999

The purpose of this paper is to expand Stowell (1993), Stowell (1995), Stowell (1996)'s syntactic analysis of tense in English. Stowell treats Tense as a dyadic predicate of temporal ordering which takes those two time-denoting phrases as its arguments. He further argues that those two morphemes 'resent' and 'past' are polarity-sensitive elements encoding an LF-scope relation with respect to true PAST tense. This paper proposes that English future 'will' should be treated as a true tense and that its future morpheme is an anti-PAST polarity item. It also provides a syntactic interpretation of a peculiar morphological aspect of English that it has no future form of the verb. To this end, Stowell's analysis is incorporated into the Minimalist program of Chomsky(1995). It is proposed that, unlike in other languages like French and Spanish, FUTURE in English is of an affix. This provides an intuitively correct description of why English verbs do not have a future form like other languages. The last but not least point which this paper will discuss is that Ogihara (1995a)'s claim that the referential theory of tensed sentences is inadequate is untenable.

• English Language & Literature Teaching
• /
• v.11 no.1
• /
• pp.245-260
• /
• 2005

Various kinds of cognitive aspects must be inclusively considered for understanding to transitivity. Basically, transitivity is not a notion of pure lexical items, pure syntactic dimensions and only semantic interpretation. That is, transitivity is affected by the extralinguistic phenomena which contain speakers psychological and cognitive interpretations. It is based on human cognitive abilities to interpret the exact meaning of language expressions. To sum up, transitivity has a relation with the observer's (speaker's) cognitive interpretation within the entire contexts.

• Korean Journal of English Language and Linguistics
• /
• v.1 no.4
• /
• pp.609-627
• /
• 2001

Pronouns are frequently used in English, and their resolution is important to capture meaning of sentences. This paper provides a computational implementation for pronoun readings in English, based on Chierchia's (1988) Binding Theory in Categorial Grammar. A CCG-like system is newly devised for implementing his ideas, where syntactic phenomena are represented by the functor-argument relations of categories. This relation triggers resolution algorithms, and reflexives and pronominals are resolved succinctly. In sum, this paper gives an efficient resolution algorithm for English pronouns within Categorial Grammar.

• Proceedings of the Korean Society for Language and Information Conference
• /
• 2008.06a
• /
• pp.138-146
• /
• 2008

Frequency adverbs can be interpreted as an adverb of quantification, and also as a frequentative adverb. These interpretations are related to the frequency adverbs' distributions, and the relation between semantics and syntax of frequency adverbs can be observed more explicitly when they appear with some other expressions in a sentence. Two frequency adverbs in Korean, cacwu and cakkwu, which seem to mean 'often/frequently', will be dealt with. We will specify their syntactic position by their interpretations derived from the relative ordering with other elements.