Songbirds as vocal learners have been one of the most popular model species to investigate the biological prerequisite to human language. Their songs consist of syllables, which appear as pulse trains in sound spectrograms. When describing the song sequence, researchers consider the syllable to be the unit of the song. Moreover, artificial grammar learning studies asking whether songbirds recognize structural regularities observed in human language often design stimuli using song syllables as components. However, whether syllables are perceptual units is yet to be determined. We found that Bengalese finches, a species of songbird, responded significantly less to one specific syllable when it was temporally placed close to the preceding syllable. The proximity, or silent interval was within the range of what is produced in the natural songs of both Bengalese and zebra finches, and what has been used in other artificial grammar learning studies using zebra finches. Our results suggest the need for a reinterpretation of the description of birdsong structure and of previous artificial grammar learning studies.

中文翻译：

---

鸣禽会一个音节一个音节地听到歌曲吗？

作为声乐学习者的鸣禽一直是研究人类语言生物学先决条件的最受欢迎的模型物种之一。他们的歌曲由音节组成，在声谱图中显示为脉冲序列。在描述歌曲序列时，研究人员将音节视为歌曲的单位。此外，询问鸣禽是否识别在人类语言中观察到的结构规律的人工语法学习研究经常使用歌曲音节作为组成部分来设计刺激。然而，音节是否是感知单位还有待确定。我们发现孟加拉雀，一种鸣禽，当一个特定的音节在时间上靠近前一个音节时，对它的反应明显减少。接近，或无声间隔在孟加拉雀和斑胸草雀的自然歌曲中产生的范围内，以及在其他使用斑胸草雀的人工语法学习研究中使用的范围内。我们的结果表明需要重新解释鸟鸣结构的描述和以前的人工语法学习研究。