Typologically-driven parallels between nominal modifiers and arguments in the clause. A formal account

Greenberg’s (1966) Universal 20 is a widely discussed typological statement that is concerned with the crosslinguistic syntactic variation of demonstratives, cardinal numerals and adjectives modifying the noun. The exact formulation of this universal hinges on the collection of data from the widest range of natural languages possible. The theoretical explanation of the observed patterns is guided instead by the presence of two extremes. First, there are many unattested permutations. Second, a fistful of highly frequent orders manage to cover a large part of the observed crosslinguistic variation. Moreover, the theoretical debate on Greenberg’s U20 turns out to be related to the variability among languages with respect to the order of the verb and its arguments in mono-transitive clauses. In fact, the existence and the nature of correlations between the nominal and the clausal domains are relevant to both syntacticians and typologists who argue that natural languages are ruled by some universal principles. In particular, the generative tradition addresses the relation between the nominal and the clausal domains as a debate on the NP- vs. DP-hypothesis (Blümel & Holler 2021). The present work will however follow Corver (2013) in arguing that such a debate can be better termed as an argument on the Lexical vs. Functional Head Hypothesis for the nominal domain (LHH vs. FHH). The present dissertation tackles Greenberg’s U20 and the potential parallels with the clausal domain by crossreferencing data that are publicly available. The generalizations that can be subsequently observed suggest that many theoretical assumptions previously made in the literature must be recast. Although the nominal and the clausal domains appear to affect the word order of one another, they do not show the parallel syntactic behaviors predicted by either the DP-/FH hypothesis or the Consistent Head Serialization stated by Primus (2001: 855). In this respect, the analysis of U20 proposed by Roberts (2017b) has a twofold merit. On the one hand, it provides some insights that set the stage for the data collection presented in this work, while on the other hand it suggests that the correlations between the nominal and the clausal domains can be captured by means of some macroparameters à la M. C. Baker (2008a). However, neither Roberts (2017b) nor any other syntactic analysis presently available manages to account for all the facts. To provide a better analysis of Greenberg’s U20, the present dissertation follows Chiang & Scheffler (2008) in adopting a computationally empowered version of Tree Adjoining Grammars (TAGs). The 2-delayed 2-Multi-Component TAG subsequently proposed manages to overcome the limits that Longenbaugh (2014) and Georgi (2019) point out with respect to the alternate TAG approach formulated by Frank (2002) et seq. By virtue of the adopted formalism, the present work not only provides an adequate account of Greenberg’s U20 that supports the FHH, but it also solves theoretical issues related to TAGs, like the missing link problem (Kallmeyer & Kuhlmann 2012, Chen-Main & Joshi 2014). By providing many insights on syntactic matters related to both the nominal and the clausal domains, 2-delayed 2-Multi-Component TAGs finally allow to set a rich research agenda for the future.