WEKO3
アイテム
文記憶のネットワーク活性化説に関するコンピュータ・シミュレーション : S-O-V型文について
https://doi.org/10.18999/bulfep.32.21
https://doi.org/10.18999/bulfep.32.21041f5acc-598c-403c-8f0a-ae8042ca6481
名前 / ファイル | ライセンス | アクション |
---|---|---|
KJ00000725979.pdf (2.1 MB)
|
|
Item type | 紀要論文 / Departmental Bulletin Paper(1) | |||||
---|---|---|---|---|---|---|
公開日 | 2006-01-06 | |||||
タイトル | ||||||
タイトル | 文記憶のネットワーク活性化説に関するコンピュータ・シミュレーション : S-O-V型文について | |||||
言語 | ja | |||||
その他のタイトル | ||||||
その他のタイトル | A COMPUTER SIMULATION FOR THE NETWORK ACTIVATION THEORY OF SENTENCE MEMORY : A Case for S-O-V sentences | |||||
言語 | en | |||||
著者 |
都築, 誉史
× 都築, 誉史× TSUZUKI, Takashi |
|||||
アクセス権 | ||||||
アクセス権 | open access | |||||
アクセス権URI | http://purl.org/coar/access_right/c_abf2 | |||||
抄録 | ||||||
内容記述タイプ | Abstract | |||||
内容記述 | The present study is designed to examine a computer simulation programed for testing the validity of the network activation theory of sentence memory. According to Hayes-Roth & Hayes-Roth (1977), basic assumptions for the network activation theory are summarized as follows. (1) Each word is represented as an unitary node in the network, and relationships between nodes are represented by appropriately labeled links that connect them. (2) The meaning of a word is derived from various relationships it has with other nodes. Semantically related words are more directly interconnected in the network than semantically unrelated words. (3) A sentence is presented as a subnetwork that includes nodes and links, and a higher-order node can represent the subnetwork connected to it. (4) A linguistic input causes activation of the network representation, and activation spreads in parallel along all the links connected to an active node. (5) The activation spreads from each activated node is limited at any given point in time. (6) The strength of each link in memory is an increasing function of the total amount and the frequency of activation it has received, and a decreasing function of the time since its last activation. For this study, the above network activation theory was translated following Miller (1981) into a LISP 1.5 computer program that consists of an activation system, encoding processes, and retrieval processes. The last two processes are carried out by the common activation system. In this model, activation originated from the active nodes spreads across links. The process of the parallel spreads of activation are simulated by computing the shortest time necessary for activation to cross a link and to active a new node. Generally, the time is computed by: Time = 1/(link strength * link activation). For the model's encoding process, a S-O-V sentence is encoded by gradually constructing a working memory structure called the Encoding Context Network based on the words of the sentence and from the nodes of the semantic network that are related to the words. For the retrieval process, the sentence is searched for along one of these Encoding Context Network structures. The model simultaneously produces estimates of the percent correct, the reaction time, and the confidence rating. These dependent variables are compared to experimental data of sentance memory. The presented simulation is to explore the extent to which the representational format in memory preserves a sentence's meaning and wording. Many researchers have shown that the memory for a sentence's meaning is far superior to the retention of that sentence's wording. Although recently, several experiments using the reaction time and the confidence rating showed that distractor sentences synonymous in meaning with previously studied sentences could successfully be discriminated even in the delayed recognition test. For example, in an experiment of Tsuzuki (1983), following an incidental study session and a five-minute intervention task, the subject was given a verification task that required responding true or false to studied sentences, synonymous distractor sentences (constructed by substituting verbs with their synonyms), and nonsynonymous distractor sentences (constructed by substituting verbs with nonsynonymous ones). In this experiments, subjects verified studied sentences more accurately, faster, and more confidently than synonymous and nonsynonymous sentences. The simulation of these experiments took place in two separate stages, the encoding process and the retrieval process. The results indicated that the corresponding estimates of the percent correct, the reaction time, and the confidence rating from the simulation agreed with those derived from the experiments. This finding provids an evidence to support the network activation theory of sentence memory. | |||||
言語 | en | |||||
内容記述 | ||||||
内容記述タイプ | Other | |||||
内容記述 | 国立情報学研究所で電子化したコンテンツを使用している。 | |||||
言語 | ja | |||||
出版者 | ||||||
出版者 | 名古屋大学教育学部 | |||||
言語 | ja | |||||
言語 | ||||||
言語 | jpn | |||||
資源タイプ | ||||||
資源タイプ識別子 | http://purl.org/coar/resource_type/c_6501 | |||||
資源タイプ | departmental bulletin paper | |||||
出版タイプ | ||||||
出版タイプ | VoR | |||||
出版タイプResource | http://purl.org/coar/version/c_970fb48d4fbd8a85 | |||||
ID登録 | ||||||
ID登録 | 10.18999/bulfep.32.21 | |||||
ID登録タイプ | JaLC | |||||
ISSN(print) | ||||||
収録物識別子タイプ | PISSN | |||||
収録物識別子 | 03874796 | |||||
書誌情報 |
ja : 名古屋大學教育學部紀要. 教育心理学科 巻 32, p. 21-47, 発行日 1985-12-20 |
|||||
フォーマット | ||||||
値 | application/pdf | |||||
著者版フラグ | ||||||
値 | publisher | |||||
URI | ||||||
識別子 | http://hdl.handle.net/2237/3687 | |||||
識別子タイプ | HDL |