ANNOTATING PROSODYWITH POLAR: CONVENTIONS FOR A DECOMPOSITIONAL ANNOTATION SYSTEM
ANNOTATING PROSODYWITH POLAR: CONVENTIONS FOR A DECOMPOSITIONAL ANNOTATION SYSTEM
复制标题
用 POLAR 注释韵律：分解注释系统的约定
DOI：
--
复制
发表时间：
2019
期刊：
影响因子：
--
通讯作者：
S. Shattuck
中科院分区：
文献类型：
--
作者：
Byron Ahn;N. Veilleux;S. Shattuck
话语的语调特征根据语言意义而变化;即使控制意义，它们也可能因说话者而异，甚至因特定说话者的话语而异。语音注释系统通常将不同的特征捆绑在一起，例如f0缩放、对齐和突出（例如，标签如H*）。这可能使得难以记录变化，并且还难以确定信号的哪些方面（i）来自音韵学-语音学界面，（ii）根据抽象语言特征而变化，（iii）取决于方言，社会背景或情绪状态，或者（iv）可能只是噪声。这促使了一个新的注释系统：点，水平和范围（PoLaR）。PoLaR从现有的理论（AM理论）和转录（IPO，IViE，RPT）系统中获得灵感（并可以与之一起使用），但既不是语音/认知模型，也不是声学/物理模型。它使用透明地对应于声学和/或母语者感知的方面的标签来隔离个体韵律特征。
An utterance’s intonational characteristics vary according to linguistic meaning; even controlling for meaning, they may vary across speakers, and even across utterances for a given speaker. Phonological annotation systems typically bundle together disparate characteristics, such as f0 scaling, alignment, and prominence (e.g., labels like H*). This can make it difficult to document variation, and also to determine which aspects of the signal (i) result from the phonology-phonetics interface, (ii) vary according to abstract linguistic features, (iii) depend on dialect, social context, or emotional state, or (iv) may be simply noise. This has motivated a new annotation system: Points, Levels, and Ranges (PoLaR). PoLaR takes inspiration from (and can be used alongside) existing theoretical (AM theory) and transcriptional (IPO, IViE, RPT) systems, but is neither a phonological/cognitive model, nor an acoustic/physical model. It isolates individual prosodic characteristics, using labels that transparently correspond to aspects of the acoustics and/or native speaker perception.