Nooteboom S G, Kruyt J G
Institute for Perception Research/IPO, Eindhoven, The Netherlands.
J Acoust Soc Am. 1987 Nov;82(5):1512-24. doi: 10.1121/1.395195.
This article reports on an experiment examining some perceptual consequences of correspondences between accent patterns, the distribution of plus and minus focus, and the distribution of new and given information in Dutch spoken sentences. "Accent patterns" refer here to the distribution of intonational accents over spoken sentences. Each accent marks a sentence constituent as plus focus, i.e., as highlighted by the speaker. Constituents not so marked are called minus focus. The main questions examined here are to what extent are plus focus constituents generally perceived as conveying new information, and minus focus constituents as conveying earlier introduced or given information. The linguistic material for the experiment was formed by brief radio news items, each two sentences long. Leading sentences determined the distribution of new and given information in target sentences. The accent patterns and, hence, the possible focus distributions in the target utterances were varied systematically by manipulating their synthetic pitch contours according to the rules for Dutch intonation. Subjects were asked to rate on a scale from 1-10 the acceptability of each possible combination of a leading with a target utterance. Results showed that the most preferred or acceptable distributions of new and given information closely match the distributions of plus and minus focus. It was also found that new information can hardly ever acceptably be associated with minus focus, but given information can rather often, although not always, acceptably be associated with plus focus. This appears to be limited to certain conditions, defined by a combination of syntactic and focus structure of the sentence. In these conditions, plus focus cannot be perceived only as signaling new information, but also as highlighting thematic relations with the context. These results are related to work on text-to-speech systems.
本文报道了一项实验,该实验研究了荷兰语口语句子中重音模式、正负焦点分布以及新旧信息分布之间对应关系的一些感知结果。这里的“重音模式”指的是语调重音在口语句子中的分布。每个重音将一个句子成分标记为正焦点,即被说话者突出强调的成分。未被如此标记的成分称为负焦点。这里研究的主要问题是,正焦点成分在多大程度上通常被视为传达新信息,而负焦点成分在多大程度上被视为传达先前引入的或已知的信息。实验所用的语言材料由简短的广播新闻条目组成,每个条目包含两个句子。引导句决定了目标句子中新旧信息的分布。通过根据荷兰语语调规则操纵目标话语的合成音高轮廓,系统地改变了重音模式,进而改变了目标话语中可能的焦点分布。要求受试者对引导句与目标话语的每种可能组合的可接受性在1至10的量表上进行评分。结果表明,新旧信息最受青睐或最可接受的分布与正负焦点的分布密切匹配。研究还发现,新信息几乎不可能与负焦点可接受地关联,但已知信息相当频繁地(尽管并非总是如此)可以与正焦点可接受地关联。这似乎仅限于某些由句子的句法和焦点结构组合所定义的条件。在这些条件下,正焦点不能仅被视为表示新信息的信号,还被视为突出与上下文的主题关系。这些结果与文本转语音系统的研究相关。