Constraints on Linearization in Nominal Modification

众所周知，在许多语言中，名词前修饰语，如用作定语的形容词，不允许在它们和中心名词之间插入其他材料。因此，“the afraid of the doctor patient”在英语中是完全不合语法的。学者们利用这一事实引入了所谓的中心语末过滤器（HFF），它禁止在名前修饰语中使用后置中心语材料。与此同时，这种约束既太强又太弱。首先，有些语言，如希腊语或俄语，实际上允许违反HFF的A-XP-N顺序，这意味着HFF过度生成。第二，有些语言，如巴斯克语，其名词后修饰语，不属于HFF的应用范围，受到类似的限制，要求它们以中心语开头，不允许镜像的N-XP-A顺序。在此基础上，我们假设这些HFF效应的基础是定语中心语的形态状态与形容词在其表语形式中所表现出的特征之间的相互作用。只有当表语形容词没有表现出DP中的所有活跃特征，并且属性标记在本质上是词缀时，才会出现HFF效应。虽然这种概括已经适用于有限的语言，但这一假设还不能完全证实。为此，需要对代表所有相关语系的语言样本进行大规模的跨语言调查，并进行理论分析，解释为什么这样的概括成立。这正是我们在这个项目中要做的。我们将研究这种概括在多大程度上成立，或者是否需要进一步完善，并解释为什么定语语素的形态状态和形容词的特征之间的界面在特定的线性化模式中起着如此重要的作用。这些研究的结果不仅会更好地理解复杂名词性表达的方式，线性化，但也创造了新的见解之间的关系形态和句法操作和功能，他们的基础。

It is a well-known observation that, in many languages, prenominal modifiers, such as attributively used adjectives, do not allow other material between them and their head nouns. Thus, 'the afraid of the doctor patient' is fully ungrammatical in English. Scholars have taken this fact to introduce the so-called Head-Final Filter (HFF) that bars post-head material in prenominal modifiers. At the same time, this constraint is both too strong and too weak. First, there are languages, like Greek or Russian, in which the HFF-violating A-XP-N order is in fact allowed, which means that the HFF overgenerates. Second, there are languages, like Basque, where postnominal modifiers, which fall outside the domain of the application of the HFF, are subject to an analogous restriction that requires them to be head-initial and disallows the mirrored N-XP-A order. In this respect, the HFF undergenerates.Basing ourselves on earlier work (Alexeyenko & Zeijlstra 2021), we hypothesize that what underlies these HFF-effects is an interaction between the morphological status of attributive heads and the features that adjectives display in their predicative forms. Only if predicative adjectives do not display all the features that are active in the DP and if the attributive marker is affixal in nature, HFF-effects emerge.While such a generalization already holds for a limited set of languages, this hypothesis cannot be fully confirmed yet. For that, a large-scale cross-linguistic survey with a language sample representing all relevant language families is needed, as is a theoretical analysis that explains why such a generalization holds. This is exactly what we will carry out in this project. We will investigate to what extent this generalization holds or whether it is in need of refinement and we will provide an explanation of why the interface between the morphological status of attributive morphemes and the features that are active on adjectives plays such a major role in licensing particular linearization patterns.The outcomes of these investigations will not only yield a better understanding of the way that complex nominal expressions can be linearized, but also create new insights with respect to the relation between morphological and syntactic operations and the features that underlie them.