• 대한음성학회지:말소리
• /
• 제51호
• /
• pp.39-55
• /
• 2004

Working memory is the term used to refer to the mental workplace in which information can be temporarily stored and manipulated during complex everyday activities such as understanding language. The studies on language and working memory are based on Baddeley's phonological working memory and Daneman and Carpenter's functional working memory. This article reviews two working memory models and the studies on language and working memory based on each model. These are described in the implication of working memory in language development and specific language impairment-evaluation and treatment.

• 말소리와 음성과학
• /
• 제3권4호
• /
• pp.95-102
• /
• 2011

The purpose of this study was to identify the characteristics of the phonological discrimination ability and phonological working memory of 10 typically developing children aged 4, and 10 other children with Specific Language Impairments whose language age is similar. In orders to compare their phonological discrimination ability among phonological awareness, discrimination tasks were conducted at the syllable and phoneme levels. Also, in order to compare their phonological working memory, the subjects repeated nonsense syllables. The research results may be summarized as follows: First, the children with Specific Language Impairments demonstrated a lower performance than the typically developing children in phonological discrimination ability at both syllable and phoneme levels, and the difference between the groups was statistically significant. Second, the children with Specific Language Impairments exhibited a lower phonological working memory performance in all syllables compared with normal children. Although there was no significant difference in 2 and 3 syllables, a significant difference appeared as the length of the syllables became longer from 4 to 6 syllables. It is deemed necessary to conduct research into qualitative and quantitative differences through an formal assessment of the phonological awareness and phonological working memory of children with Specific Language Impairments.

• 대한전기학회:학술대회논문집
• /
• 대한전기학회 2005년도 학술대회 논문집 정보 및 제어부문
• /
• pp.201-203
• /
• 2005

The best method to study language is to talking with a native speaker. A voice recognition technology can be used to develope a language study machine. SD(Speaker dependant) and SI(speaker independant) voice recognition method is used for the language study machine. MP3 Player. FM Radio. Alarm clock functions are added to enhance the value of the product. The machine is designed with a DSP(Digital Signal Processing) chip for voice recognition. MP3 encoder/decoder chip. FM tumer and SD flash memory card. This paper deals with the application of SD ad SD voice recognition. flash memory file system. PC download function using USB ports, English conversation text function by the use of SD flash memory. LCD display control. MP3 encoding and decoding, etc. The study contents are saved in SD flash memory. This machine can be helpful from child to adult by changing the SD flash memory.

• 음성과학
• /
• 제8권4호
• /
• pp.119-129
• /
• 2001

Language is one aspect of cognition, along with attention and concentration, learning and memory, visuospatial abilities, and executive function. The purpose of this study was to determine the effect of language intervention by cognitive approach on language expressive performance in a patient with conduction aphasia. This study used several tasks such as Attention and concentration task, visual memory tasks, memory tasks, categorization, divergent thinking, self-monitoring and evaluate thinking. The effects of treatment were evaluated by periodic probing of both trained and untrained familiar words in three tasks; picture naming, answering to questions and telling stories. The results showed improvements both in trained and untrained words. Therefore, we concluded that expressive language performance of this aphasic patient is amenable to this intervention, and that cognitive therapy approach can be useful.

• 인지과학
• /
• 제23권4호
• /
• pp.425-455
• /
• 2012

다차원 구어 단기기억 기제는 크게 음운적 통로와 어휘-의미적 통로로 구분된다. 전자를 음운단기기억, 후자를 의미단기기억이라고 할 수 있는데, 단기기억 과제를 정상적으로 수행하기 위해서는 두 통로 모두의 정보를 활발히 활용하여야 한다. 그리고 음운단기기억은 다시 음운입력완충기와 음운출력완충기로 나누어지며, 음운입력완충기는 음운자극의 입력 시, 음운출력완충기는 음운 산출 시에 작동한다. 본 연구에서는 유사한 수준의 전도 실어증 증상을 보이는 세 명의 환자에 대해, 각각의 언어 수행력을 구어 단기기억의 다차원적 측면에서 분석하였다. 그러기 위하여 세 명의 전도 실어증 환자들에게 단어 수준과 문장 수준에서 스스로 말하기, 따라말하기, 스스로 쓰기, 받아쓰기의 네 가지 양태의 언어과제를 실시하여 수행력을 비교 분석하였고, 숫자폭검사와 언어학습검사를 이용하여 음운단기기억력과 의미단기기억력을 평가하였다. 그 결과 세 대상자들은 네 양태의 언어 검사에서 다양한 수행력과 오반응 유형을 보였고, 단기기억력 검사 결과도 동일하게 나타나지 않았다. 즉 전도 실어증 환자들의 언어 수행력은 의미단기기억 또는 음운단기기억의 결함으로 설명될 수 있으며, 음운단기기억 가운데에서도 음운입력완충기, 음운출력완충기 혹은 둘 다의 결함 여부에 따라 언어특성이 상이하게 나타날 가능성을 제시하고 있다. 본 연구에서는 전도 실어증 환자들의 언어 검사와 단기기억력 검사 결과를 바탕으로, 언어와 다차원 구어 단기기억력과의 관계에 대하여 논의하고 있다.

• International Journal of Knowledge Content Development & Technology
• /
• 제5권2호
• /
• pp.75-101
• /
• 2015

In this paper, we investigate the notion of error correction memory applied to technical texts. The main purpose is to introduce flexibility and context sensitivity in the detection and the correction of errors related to Constrained Natural Language (CNL) principles. This is realized by enhancing error detection paired with relatively generic correction patterns and contextual correction recommendations. Patterns are induced from previous corrections made by technical writers for a given type of text. The impact of such an error correction memory is also investigated from the point of view of the technical writer's cognitive activity. The notion of error correction memory is developed within the framework of the LELIE project an experiment is carried out on the case of fuzzy lexical items and negation, which are both major problems in technical writing. Language processing and knowledge representation aspects are developed together with evaluation directions.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제15권3호
• /
• pp.837-852
• /
• 2021

Recurrent neural network (RNN) architectures have been used for language modeling (LM) tasks that require learning long-range word or character sequences. However, the RNN architecture is still suffered from unstable gradients on long-range sequences. To address the issue of long-range sequences, an attention mechanism has been used, showing state-of-the-art (SOTA) performance in all LM tasks. A differentiable neural computer (DNC) is a deep learning architecture using an attention mechanism. The DNC architecture is a neural network augmented with a content-addressable external memory. However, in the write operation, some information unrelated to the input word remains in memory. Moreover, DNCs have been found to perform poorly with low numbers of weight parameters. Therefore, we propose a robust memory deallocation method using a limited retention vector. The limited retention vector determines whether the network increases or decreases its usage of information in external memory according to a threshold. We experimentally evaluate the robustness of a DNC implementing the proposed approach according to the size of the controller and external memory on the enwik8 LM task. When we decreased the number of weight parameters by 32.47%, the proposed DNC showed a low bits-per-character (BPC) degradation of 4.30%, demonstrating the effectiveness of our approach in language modeling tasks.

• International Journal of Internet, Broadcasting and Communication
• /
• 제12권2호
• /
• pp.45-50
• /
• 2020

A Language model with neural networks commonly trained with likelihood loss. Such that the model can learn the sequence of human text. State-of-the-art results achieved in various language generation tasks, e.g., text summarization, dialogue response generation, and text generation, by utilizing the language model's next token output probabilities. Monotonous and boring outputs are a well-known problem of this model, yet only a few solutions proposed to address this problem. Several decoding techniques proposed to suppress repetitive tokens. Unlikelihood training approached this problem by penalizing candidate tokens probabilities if the tokens already seen in previous steps. While the method successfully showed a less repetitive generated token, the method has a large memory consumption because of the training need a big vocabulary size. We effectively reduced memory footprint by encoding words as sequences of subword units. Finally, we report competitive results with token level unlikelihood training in several automatic evaluations compared to the previous work.

• 한국융합학회논문지
• /
• 제10권9호
• /
• pp.167-172
• /
• 2019

본 연구는 초기 치매환자를 대상으로 전산화 인지재활프로그램이 전두엽 집행기능 중 생성 이름대기에 미치는 효과를 파악하고, 치매 환자의 조기 언어중재에 관한 기초자료를 제공하였다. 분석 대상은 경증치매환자 29명으로서 전산화 인지재활프로그램(CoTras)을 중재하는 실험군 21명과 지필과 테이블활동 등 전통적인 면대면 언어재활을 중재하는 대조군 8명으로 분류되었다. 실험집단과 대조집단은 모두 12주 동안 순차적 언어 회상 기억훈련, 연합회상 기억훈련, 언어 범주화 기억훈련, 언어통합 기억훈련을 진행하였다. 의미유창성과 음소유창성의 사전검사점수를 공변량으로 통제한 Welch's robust ANCOVA분석 결과, 실험집단과 대조집단은 의미유창성과 MMSE-K의 변화에서 유의미한 차이가 있었다(p<0.05). 반면에, 음소유창성은 실험군과 대조군 모두 기초선단계에 비해서 12주간의 치료 후 유의미하게 증가하였지만 통계적으로 치료방법에 따른 유의미한 차이는 없었다. 본 연구의 결과는 전산화 인지재활프로그램이 전통적인 인지재활에 비해서 의미유창성의 향상에 효과적일 가능성을 시사한다.

• 한국정보과학회 언어공학연구회:학술대회논문집(한글 및 한국어 정보처리)
• /
• 한국정보과학회언어공학연구회 2018년도 제30회 한글 및 한국어 정보처리 학술대회
• /
• pp.31-34
• /
• 2018

대화 시스템에서 사람과 기계와의 모든 발화에서 발생하는 상황들을 모두 규칙화할 수 없기 때문에 자연스러운 대화가 단절되는 breakdown 현상이 빈번하게 일어날 수 있다. 이런 현상이 발생하는 이유는 다음과 같다. 첫째, 대화에서는 다양한 도메인이 등장하기 때문에 시스템이 커버할 수 있는 리소스가 부족하며, 둘째, 대화 데이터에서 학습을 위한 annotation되어 있는 많은 양의 코퍼스를 보유하기에는 한계가 있으며, 모델에 모든 대화 흐름의 히스토리를 반영하기 어렵다. 이런 한계점이 존재함에도 breakdown detection은 자연스러운 대화 시스템을 위해서는 필수적인 기능이다. 본 논문은 이런 이슈들을 해소하기 위해서 memory attention기반의 새로운 모델을 제안하였다. 제안한 모델은 대화내에 발화에 대해 memory attention을 이용하여 과거 히스토리가 반영되기 때문에 자연스러운 대화흐름을 잘 detection할 수 있으며, 기존 모델과의 성능비교에서 state-of-the art 결과를 도출하였다.