研究課題/領域番号 |
20K13055
|
研究機関 | 名古屋商科大学 |
研究代表者 |
|
研究期間 (年度) |
2020-04-01 – 2024-03-31
|
キーワード | Iconicity / Predictablity / Machine Learning / English |
研究実績の概要 |
Natural language processing is a field of study that is interested in teaching artificial intelligence how to use language the same way that humans do. This research involves teaching artifical intelligence how to understand sound symbolism. Sound symbolism is the non-arbitrary relationship between sounds and meaning. It is stochastic in nature, in other words, it can be measure through probability but does not apply to all cases. My research has shown that artificial intelligence is particularly good at sound symbolism, even when the AI has very few langauge sample to learn from. By teaching the AI how to understand sound symbolism, we learn how humans percieve speech sounds and how probability is incorporated into the phenomenon.
There are currently two articles that are about to be submitted. One of the articles was previously submitted, but was rejected on the basis of the data treatment. That treatment has been reconducted and the study will be resubmitted. Additionally, I am about to travel for a conference to present my findings. My research continues in this manner until the corona virus issue is resolved.
|
現在までの達成度 (区分) |
現在までの達成度 (区分)
3: やや遅れている
理由
Project is slightly delayed due to the coronavirus and a recent project rejection. However, the research is going reasonably well and I expect to have two publications this year.
|
今後の研究の推進方策 |
My research primarily examines the role of probabilty on speech perception, however, my typical psycholinguistic experiments are unsafe during a pandemic. This is the reason why my methodology involves AI, rather than traditional psycholinguistic experiments. Once the threat of the coronavirus is reduced, I will return to more traditional methods and my research funding will be used in the recruitment of participants. Otherwise, I will continue running experiments on existing corpus data.
|
次年度使用額が生じた理由 |
Due to Coronavirus, in-person experiments have not been conducted. Therefore, less of the budget has been spent than expected. In the future, funds will be used to purchase access to existinng speech corpora, software for running AI, computer equiptment, and, if the impact of the corona virus is reduced, paying participants.
|