Researchers Decode Elephant Communication

研究人员破译大象通讯

2021-07-14 14:17 multilingual

本文共332个字,阅读需4分钟

阅读模式 切换至中文

Lauded by Scientific American as the “first ‘Google Translate’ for elephants,” a team of researchers at the non-profit animal conservation organization ElephantVoices has developed a digitized collection of sounds and behaviors decoding methods of communication used by elephants. The program, also called the Elephant Ethogram, includes a collection of more than 500 behaviors and almost 3,000 videos, photographs, and audio files of elephant vocalizations. Elephants are well-known within the animal kingdom for having a highly complex form of communication compared to other non-human animals, however this research project — which is ongoing — is the first digitized collection of elephant behaviors and vocalizations. “At a time when biodiversity is plummeting and the lives of elephants are being heavily impacted by humans, we also want to spell out to the world what we stand to lose,” Joyce Poole, co-founder of ElephantVoices, told Scientific American. Animal communication remains a topic of debate in linguistics and other fields — the question of whether or not non-humans have communication systems that can be considered language is fairly controversial. Earlier this year, MultiLingual reported on the Cetacean Translation Initiative (CETI), a project spearheaded by a team of researchers at Texas A&M who hope to use artificial intelligence to “translate” the vocalizations of sperm whales. The Elephant Ethogram project is a bit different from CETI — it’s not AI-driven, however it does consist of a large pool of digitized data, somewhat resembling the corpora often used in machine translation and corpus linguistics. An ethogram is a collection of behaviors observed in a certain species of animals — historically, these have largely included written observations and photographs of behavior, however technological advancements have allowed scientists to collect video and audio records in addition to these written descriptions. The recently developed ethogram serves to shed light on elephant behaviors and vocalizations in an accessible fashion, allowing human researchers to easily decipher (or perhaps, “translate”) communication between elephants, making it a valuable step toward better understanding non-human communication systems.
非营利性动物保护组织“大象之声”的一个研究小组开发了一套数字化的声音和行为解码方法,被《科学美国人》誉为“大象的第一个‘谷歌翻译’”。该程序也被称为大象行为图,包括500多个行为和近3000个视频,照片,以及大象发声的音频文件。 与其他非人类动物相比,大象在动物界以其高度复杂的交流方式而闻名,然而这项正在进行中的研究项目是首次将大象的行为和发声进行数字化收集。“大象之声”联合创始人乔伊斯·普尔在接受《科学美国人》采访时说:“在生物多样性急剧减少,大象的生活受到人类严重影响的时候,我们也想向全世界说明我们将失去什么。” 动物交流在语言学和其他领域仍然是一个争论的话题--非人类是否有可以被认为是语言的交流系统的问题是相当有争议的。今年早些时候,《多语言》报道了鲸类翻译倡议(CETI),这是一个由德克萨斯农工大学的一个研究团队牵头的项目,他们希望利用人工智能来“翻译”抹香鲸的发声。 Elephant Ethogram项目与CETI有点不同--它不是由人工智能驱动的,但它确实包含大量数字化数据,有点像机器翻译和语料库语言学中经常使用的语料库。动物行为图是在某一物种中观察到的行为的集合--历史上,这些行为图主要包括行为的书面观察和照片,然而技术的进步使得科学家们除了这些书面描述之外,还可以收集视频和音频记录。 最近开发的行为特征图以一种易于理解的方式揭示了大象的行为和发声,使人类研究人员能够轻松地破译(或者“翻译”)大象之间的交流,从而使它朝着更好地理解非人类交流系统迈出了有价值的一步。

以上中文文本为机器翻译,存在不同程度偏差和错误,请理解并参考英文原文阅读。

阅读原文