Earticle

This paper addresses the critical need for sign language recognition to improve communication for individuals with hearing impairments. Our approach involves leveraging the capabilities of computer vision and deep learning to develop an efficient sign language recognition system. By utilizing the Mediapipe library, we extract detailed hand landmarks from video input, thereby effectively capturing the nuanced movements and configurations inherent in sign language gestures. At the core of our model is the GRU network, a type of recurrent neural network (RNN) specifically designed for analyzing sequential data. The GRU architecture excels at capturing temporal dependencies, making it a well-suited choice for recognizing the dynamic nature of sign language expressions. To train our GRU network, we leverage the LSA64 dataset, which comprises comprehensive videos featuring expert sign language users. We extract hand landmarks from these videos using the Mediapipe module, enabling our model to effectively learn and recognize the intricate patterns of sign language gestures.

This paper examines the phenomenon of the influent use of Korean Sign Language based on the 40-minute discourse data of 4 Korean sign language users. The data analysis found fillers, repetition, pauses, repair, and editing expressions, which are consistent with those suggested in previous studies on spoken languages. The influent used expressions related to non-manual markers and the use of both the dominant and non-dominant factors are also observed in this study. These factors are the sign language-specific influency phenomenon, which do not appear in spoken languages. In addition, this study shows that ‘incompletion’ appears frequently in Korean Sign Language, which is different from previous reports that ‘incompletion’ is usually found even less than ‘fillers’ in spoken languages. Concerning the discourse functions of influent expressions, there are buying time, remembering, maintaining the right to speak, reviewing one’s own speech, checking the partner’s reaction, and meta-cognitive confirmation. It seems that they are all related to the importance of interaction and the efficient transmission of information between signers.

본 연구의 목적은 수어 형태소의 정의를 확인하고, 이를 바탕으로 『한국수어사전』에 나타난 수어 형태소를 식별하고 분류하는 데에 있다. 구어와 마찬가지로, 수어 연구에서도 형태소는 단어형성론뿐 아니라 음운론, 통사론 등 수어의 언어학적 영역 전반에 관여하는 주요 개념이다. 그러나 한국 수어 연구에서 수어 형태소는 보조적인 위치에 머물렀다는 점에서 아쉬움이 남는다. 본 연구에서는 구어의 형태소 정의를 참조하여 수어 형태소를 “뜻을 가진 수어소의 연결체 중 최소의 단위”로 정의하고, 계열관계와 통합관계, 그리고 문장에서의 기능을 기준으로 『한국수어사전』에 등재된 3,826개 어휘로부터 수어 형태소를 식별하였다. 식별된 형태소는 세 가지 기준으로 분류하였다. 첫 번째 기준인 자립성은 형태소가 문장 내에서 독립하여 하나의 단어로 쓰일 수 있는 성질을 말한다. 두 번째 기준인 가변성은 동일한 형태소가 형태·통사적 환경에 따라 달리 나타나는 성질을 말하며, 마지막 기준인 계열성은 여러 단어의 구성요소로 공유되는 형태소의 성질을 의미한다. 이들 세 가지 기준을 토대로 식별된 수어 형태소를 분류한 결과, 자립-가변 형태소, 자립-불변 형태소, 의존-불변-계열 형태소, 의존-불변-비계열 형태소라는 네 가지 유형이 나타나는 것을 확인하였다.

This paper aims to inspect definition of morphemes in Korean Sign Language(KSL), identify and classify sign-morpheme in 『Dictionary of Korean Sign Language』. Sign-morpheme is the major concept involved in the linguistic researches, not only word-formation, but phonology and syntax. But sign-morpheme has been recognized as an secondary concept in the previous researches. This paper defines sign-morpheme as “the smallest meaningful units with a sequence of cheremes” in reference to studies of spoken language, and identify 3,826 listed-words by paradigmatic relation, syntagmatic relation, and function in sentences. Sign-morpheme is classified according to three criteria. The first criterion, Independence, is a characteristic of morphemes that can be used independently in sentences. The second criterion is Morphological Variability which means a characteristic of morphemes that changes form depending on the morpho- syntactic environment. Paradigmaticity, the last criterion, means a characteristic of morphemes that shared in several words. Based on these criteria, sign-morpheme were classified into four types; free-variable, free- invariable, bound-invariable-paradigmatic, bound-invariable-nonparadigmatic.

Nam Ki-Hyun, 2021. A study on classification of idiomatic phrases and polysemous signs in Korean Sign Language?Focused on the [GANGHADA]-. Korean Semantics, 74. This study proposes the necessity of division between idiomatic phrases and polysemic signs in Koran Sign Language(KSL). We recognize as a problem that, in the meaning review of KSL idiomatic phrases, dependence on correspondent expressions of Korean language was traditionally found high. We discovered individual meanings for [GANGHADA] in the context. As a result, [GANGHADA] possesses five semes: 'hard(robust)', ‘strong’, 'rare occurrence of change of situation', 'no problem(confident)', and 'mentally strong.' Also the phrases [GOJIB] [GANGHADA] ‘stubborn’ have the definition of 'not yield one's ideas or will.' To judge [GANGHADA] as either polysemic sign or homographic sign, etymology was employed as the standard. In KSL there are found many signs that are generated based upon iconicity. If target signs have common origin of iconicity, they can be judged as having the same root. However, in judgment of homographic signs, [GANGHADA] has the same manuals but different non-manuals. So they can be seen as separate lexemes with different origins. However, The non-manual element of [GANGHADA] should be dealt with at a pragmatic level by revealing the speaker's feelings, feelings, and thoughts when expressing [GANGHADA] rather than functioning at the word level. Consequently [GANGHADA] shares both the semes of concrete strength and abstract strength. Common attributes of 'strong' exist for these semes; they can be seen as polysemic signs.