Thorsen N G
J Acoust Soc Am. 1980 Mar;67(3):1014-30. doi: 10.1121/1.384069.
Ten subjects identified 15 Danish utterances by a human speaker, differing only in their fundamental frequency (F0) course, as being either declarative, nonfinal, or interrogative (forced choice). Responses are closely correlated with F0: the most steeply falling intonation contours are identified as being declarative, the least falling ones as being interrogative, and contours in the middle of the continuum as being nonfinal. Several mutually interdependent parameters in the F0 course may account for the results, the two most powerful one, however, being the levels of the last stressed and the succeeding unstressed syllable, respectively, in the utterance. In a subsequent experiment, seven subjects identified the same utterances as being either declarative or nondeclarative. The majority of the (formerly) nonfinal sentences were now labeled nondeclarative, rather than being equally distributed among the declarative and nondeclarative categories. When a subset of the same utterances were multilated, identification deteriorated almost progressively with the number of syllables being cut away from the end of the utterance, but only slightly so until nothing but the first stress group remained; whereas, syllables cut away from the beginning hardly affected identification at all.
十名受试者将一位人类说话者说出的15个丹麦语语句识别为陈述句、非终句或疑问句（强制选择），这些语句仅在其基频（F0）变化上有所不同。反应与F0密切相关：下降最陡的语调轮廓被识别为陈述句，下降最少的为疑问句，而处于连续体中间的轮廓为非终句。F0变化过程中的几个相互依存的参数可能解释了这些结果，然而，其中最具影响力的两个参数分别是语句中最后一个重读音节和后续非重读音节的音高。在随后的一项实验中，七名受试者将相同的语句识别为陈述句或非陈述句。大多数（之前的）非终句现在被标记为非陈述句，而不是在陈述句和非陈述句类别中平均分布。当同一组语句的一个子集被删节时，识别能力几乎随着从语句末尾去掉的音节数量而逐渐下降，但在只剩下第一个重音组之前下降幅度很小；而从开头去掉的音节几乎对识别没有影响。
