Abstract Previous studies found that English figurative idioms alliterate at above chance levels. To permit estimation of amounts of surplus alliteration Gries (2011) calculated baseline levels using an analytic method. This article reports a follow-on investigation covering types of multiword unit (MWU) and types of interword, intraMWU phonological similarity (PhS) considered neither by Gries nor by an even earlier study. In contrast to Gries (2011), baseline levels of PhS were estimated using a stochastic method. In samples of figurative idioms upward departures from baseline levels – expressed as standardized effect sizes – ranged from small to medium for assonance, up to large for alliteration, and even larger for rhyme and alliteration-with-assonance. For samples of (relatively) non-idiomatic MWUs upward departures from baselines were generally small or, in the case of academic collocations, downward. The practicality of the stochastic method is discussed, as are a possible role of interword PhS in the conventionalization of word strings and possible roles of interword, intra-idiom PhS in oral communication. Overall, the findings are problematic for a non-usage-based theory but compatible with a cognitive linguistic theory in which motivation can operate entirely within the phonological pole of a MWU.

中文翻译：

---

英语多词单元中的多余词间语音相似度

摘要先前的研究发现，英语比喻成语在高于偶然性的水平上也存在。为了允许估计剩余的头等量，Gries（2011）使用一种分析方法计算了基准水平。本文报告了后续调查，涵盖了多字单元（MWU）和字间类型，MWU内部语音相似度（PhS）的类型，格里斯和更早的研究都没有考虑。与Gries（2011）相比，PhS的基线水平是使用随机方法估算的。在比喻习语的样本中，基线水平的向上偏离（表示为标准化效果大小）从小到中等不等，引起共鸣，从大到大不等，而对于韵律和不规则声则更大。对于（相对）非惯用MWU的样本，与基线的向上偏离通常很小，在学术搭配的情况下则向下偏离。讨论了随机方法的实用性，以及单词间PhS在常规化字符串中的可能作用以及单词间内成语PhS在口语交流中的可能作用。总体而言，这些发现对于基于非用法的理论是有问题的，但与认知语言学理论是兼容的，在认知语言学理论中，动机可以完全在MWU的语音极点内起作用。