Berent Iris, Bat-El Outi, Vaknin-Nusbaum Vered
Northeastern University, United States.
Tel-Aviv University, Israel.
Cognition. 2017 Apr;161:117-128. doi: 10.1016/j.cognition.2017.01.011. Epub 2017 Feb 7.
Duality of patterning, is, by hypothesis, a universal design feature of language. Every language constructs words from meaningful units (morphemes), which, in turn, are comprised of meaningless phonological elements (e.g., segments, syllables). But whether the language faculty does, in fact, include a separate morphological level, distinct from the phonology, is a matter of controversy. To elucidate the role of morphology, here we ask whether morphological forms are constrained by putatively universal combinatorial principles, distinct from those applying to phonological patterns. Our research exploits the structural ambiguity of doubling. Doubling (e.g., trafraf) is open to two competing interpretations-as either a purely phonological form, or as a complex morphological structure that is systematically linked to meaning (e.g., trafraf is the diminutive of traf). Our experiments show that responses to doubling (trafraf) shift radically, depending on its level of analysis. Viewed as a meaningless phonological form, doubling is dispreferred irrespective of its kind (i.e., trafraf is as bad as traftaf, even though the latter violates a morphological constraint on contiguity). But once doubling is systematically linked to meaning (i.e., as a morphological structure), the doubling dislike shifts into a reliable preference, and an additional constraint on its contiguity arises (i.e., trafraf>traftaf). Remarkably, the dissociation between morphological and phonological doubling emerges regardless of whether morphological reduplication is abundant in participants' language (in Hebrew) or relatively rare (in English). These results suggest the existence of distinct linguistic constraints that preferentially target the morphological vs. phonological levels. We discuss various explanations for the origins of these restrictions.
模式的二元性，根据假设，是语言的一种普遍设计特征。每种语言都由有意义的单位（语素）构建单词，而这些语素又由无意义的音系元素（例如，音段、音节）组成。但是语言官能是否实际上包括一个与音系学不同的独立形态学层面，这是一个有争议的问题。为了阐明形态学的作用，我们在此询问形态形式是否受到与应用于音系模式的原则不同的假定普遍组合原则的约束。我们的研究利用了重叠的结构歧义性。重叠（例如，trafraf）有两种相互竞争的解释——要么是纯粹的音系形式，要么是与意义系统相关的复杂形态结构（例如，trafraf是traf的指小词）。我们的实验表明，对重叠（trafraf）的反应会根据其分析层面而发生根本性变化。被视为无意义的音系形式时，无论重叠的类型如何，重叠形式都不受青睐（即，trafraf和traftaf一样糟糕，尽管后者违反了关于邻接性的形态约束）。但是一旦重叠与意义系统相关联（即作为一种形态结构），对重叠的厌恶就会转变为可靠的偏好，并且会出现对其邻接性的额外约束（即，trafraf>traftaf）。值得注意的是，无论形态重叠在参与者的语言中是否丰富（在希伯来语中）或相对较少（在英语中），形态和音系重叠之间的分离都会出现。这些结果表明存在不同的语言约束，它们优先针对形态学层面和音系学层面。我们讨论了这些限制起源的各种解释。
